SlideShare a Scribd company logo
1 of 34
How to regulate
foundation models:
can we do better
than the EU AI Act?
Lilian Edwards
Professor of Law,
Newcastle University
@lilianedwards
Lilian.edwards@ncl.ac.uk
April 2023
What are large or “foundation” models?
• GPT-2/3/3.5/4 (Open
AI/Microsoft)(prompt to text)(2019
on)
• “Large Language Model” or LLM
• ChatGPT
• DALL-E 2 (text to images – Google)
• Stable Diffusion (open source – text
to image)
• HarmonAI – makes AI generated
music (Stability)
• CoPilot (prompt generates computer
code – GitHub/OpenAI)
• Meta Make-me-A-Video (text to
video - Meta)
• ERNIE ( Baidu, China) (prompt to
text)
eg Stable Diffusion : img to img (open source third party code)
nurse
Stable Diffusion
doctor
DALL-E 2
ChatGPT
December
2022
Ecology of downstream deployers
Integration into search, Feb 2023
New York Times
Important (for law) features of large or
“foundation” models
• Generative – create text, images etc rather than merely
classifying or predicting (ML)
• Trained on unprecedentedly large datasets
• Often scraped from “public” Internet
• Impossible to manually review legality, privacy or harm
of every item in datasets
• Computationally expensive and retraining slow ->
• large tech co dominance
• GPT-4 training cost >$100mn
• environmentally worrying
• Training sets allow the model to assess probability of
next word, pixel etc – not direct copying
• Models are general , can have multiple uses, eg to write
a party invite, a racist attack or provide customer
support within an automated hiring system
• Generated content increasingly difficult to distinguish
from human-created content (disinfo, deepfakes)
• Outputs may be “hallucinations” Hoppner, 2023
Issues with large models
PHASE 1 STOCHASTIC PARROTS
• Don’t actually understand , just “parrot”
• Bias, discrimination, misrepresentation and stereotyping of
groups; hate speech
PHASE 2 WILL NO-ONE THINK OF THE ARTISTS?
• Image and video deepfakes
• Pastiche
• Copyright
PHASE 3 FAKE NEWS ON STEROIDS
• Fake news and “hallucination” (text + images]
• Education & plagiarism
• Digital Services Act
PHASE 4 – YOU HAVE ZERO PRIVACY, GET OVER IT
• GDPR
Solution 1 : the EU AI Act and “GPAI”
AIA “risk based” approach
• Unacceptable risk – ‘Complete’
prohibition, 4 examples – Article 5
• High-risk –Fixed categories of
risky domains, based on intended
use ; “essential requirements”
including dataset quality, human
oversight –
• Limited risk – Transparency
obligations for a few AI systems
(chatbots, deepfakes, emotion ID,
biometric categorisation) – Article
52
• Minimal risk – Codes of conduct –
Article 69
Photo Source: European Commission, Digital Strategy Website
https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai
Annex II - Products
• Machinery
• Toys
• Recreational craft and watercraft
• Lifts
• Equipment and protective systems
intended for use in potentially
explosive atmospheres
• Radio equipment
• Pressure equipment
• Cableway installations
• PPE
• Medical devices
• [...]
High-Risk AI systems (Designation) - Annexes II&III
• Annex III - Services
• Biometric identification and
categorisation of natural persons;
• Management and operation of critical
infrastructure;
• Education and vocational training;
• Employment, workers management and
access to self-employment;
• Access to and enjoyment of essential
private services and public services and
benefit;
• Law enforcement;
• Migration, asylum and border control
management;
• Administration of justice and democratic
processes
• Compliance with requirements (Art.8);
• Risk management system (Art.9);
• Data and data governance (Art.10); (“data quality”)
• “Training, validation and testing data sets shall be relevant, representative, free of errors and complete. They shall have the appropriate
statistical properties, including, where applicable, as regards the persons or groups of persons on which the high-risk AI system is
intended to be used.”
• Technical documentation (Art.11);
• Record-keeping (Art.12);
• Transparency and provision of information to users (Art.13);
• Human oversight (Art.14);
• “Human oversight shall aim at preventing or minimising the risks to health, safety or fundamental rights that may emerge when a high-
risk AI system is used in accordance with its intended purpose or under conditions of reasonably foreseeable misuse”
• Accuracy, robustness and cybersecurity (Art.15).
High-Risk AI systems (Requirements) – Arts. 8 - 15
Definition of GPAI in EU AIA
LGAIMs = Large General AI Models – Hacker, Engel and Mauer “Regulating ChatGPT and other Large generative AI Models”, February
2023
Issues : over-inclusive (no emphasis generality in “ability, task or output”); important for classification ->
Council position
‘general purpose AI system’ means an AI system that is trained on broad data at scale, is
designed for generality of output, and can be adapted to a wide range of tasks
EU Parliament, March 2023
Developers v deployers
• Developers?
• Akin to manufacturers – control
and have knowledge of the
training sets, weights, algorithms,
content moderation etc (esp if
closed source)
• Most high-risk obligations arise at
development stage (training &
human fine-tuning stages)
• Have practical power and
economic gains
• But can they handle
unforeseeable uses/risks? Or can
only tech giants  competition
issues? Open source providers?
• Deployers? [NB “users” in AIA!]
• Originally only duties on deployer
if they make a “substantial
modification” to the AI system, ie
become new provider
• Practicality ?: May be impossible
for them to fix or even audit issues
of data quality etc without access
to upstream source code, training
datasets etc (often
secret/proprietary – eg as with
GPT-3)
• “AI as a service” API/cloud model
will be prevalent till models
smaller
“Pick n Mix?”
Solutions in AIA process?
• 13 May 2022, French presidency added amendment excluding GPAIs from AIA
• Back and forth..
• Council position, Dec 2022, arts 4a-4c
• GPAI deemed high risk if they “may be used as high risk AI systems or as
components of high risk AI systems”
• Unless “explicitly excluded all high-risk uses”, but not if not in “good faith”
• European Parliament
• All generative AI models to go into high risk if
• they generate text that might be mistaken for human
• And all deepfakes and AV content showing something that “never
happened” unless an “obvious artistic work”
• & Providers owe cooperation & transparency obligations to downstream
users
• Commission to tweak the high-risk obligations by delegated Acts..
Foundation models v GPAI?? 19/4/23
“Foundation models”
• “an AI system model that is trained on broad data
at scale, is designed for generality of output, and
can be adapted to a wide range of distinctive
tasks.”
• Eg Chat GPT , Stable Diffusion
• “Trained on data scraped from entire Internet”?
Not just if labelled
• Stricter obligations - High risk ++?
• Adds sustainability obligations + independent
expert ex ante oversight
• Documented analysis and testing throughout
lifecycle
• Disclosures re copyright in training set; filters to
avoid delivery unlawful content
“GPAI”
• “AI system that can be used in and adapted to
a wide range of applications for which it was
not intentionally and specifically designed.”
• ?? E.g. “unlabelled data that need further
training by the provider, such as algorithms
developed to recognise skin cancer”
• Laxer regime, since obligations fall on high risk
providers who build on the data (?)
Deployers & value chain
• “non-binding standard contractual clauses that
regulate rights and obligations consistent with each
party’s level of control”
• Discrimination/ equality law
• Liability (product liability for AI)
• Copyright
• Content, hate speech, libel, fake
news (DSA)
• Privacy & data protection
• Personality & image rights
• Competition law
• See Hoppner
https://papers.ssrn.com/sol3/paper
s.cfm?abstract_id=4371681
• Apart from last though, only AIA
and DMA allows for structural ex
ante regulation
• Private ordering?
Solution 2 : everything but the AI Act!
1. Copyright & AI generated art: another
entire talk
• WaPost : “He used AI to win
a fine-arts competition. Was
it cheating?
• One judge said the striking
piece evoked Renaissance
art. But some critics
compared it to ‘entering a
marathon and driving a
Lamborghini to the finish
line.’”
• Effect on original artists? A
tool or a replacement??
• Boris Eldagsen The Electrician
• “winner” Sony World
Photography Awards,
• 17 April 2023
AI generated art and
copyright
• Often copyright in the art used as training dataset
inputs (eg Rutowski) – but they are not part of the
model
• No direct copying though sometimes a perfect copy
might emerge (memorisation)
• Is there actual copying of inputs?
• Is there (US) fair use or (UK/DSA) research or TDM
exception?
• Who owns the outputs?
• Again almost never direct copies
• “After the style of”..
• Derivative works?
• Some partial solutions : opt-out; haveIbeentrained ;
license, royalties or benefit sharing
• Litigation!
AI art litigation (US)
• Getty v Stability, based on copying of input works copyright Getty
• Transformative fair use?
• Anderson vs Stability & Deviant Art
• Aims to acquire rights over all OUTPUTS as derivative works from artistic works in
training set
• Defendants case to dismiss -
• NB Stability is open source, so you can analyse the underlying training sets
(cf Open (sic) AI & GPT-4)
If you thought artists were
p*ssed off by genAI..
… try the music
industry!
Can GPAI be compliant with GDPR?
• AIA high risk data quality and
transparency requirements actually
say nothing about privacy of users
• Machine learning process long
regarded as dubious but
• no federal DP law in US
• no quality of privacy,
confidentiality to data made
public in US
• GPAIs use permissionless public
data (eg Common Crawl)
• But - Replika decision, Italy, 2/2/23
• Primarily about exposure of children
to unsuitable sexualized material
• Unable to make valid contracts
Italy DPA Garantie vs ChatGPT, 31 March 2023
“
“
Open AI’s Italian Job(s)
What next?
• “reports of a 400% surge in VPN
downloads in Italy” 
• Spain, France, pan-European investigation
by EDPB
• Canada investigation
• Does GPAI fundamentally challenge the
GDPR?
• If so, which gives?? (and does it take ML
with it?)
• The end of the data Wild West?
• But are the privacy regulators really the
right ones to take generative AI on?
• (and will the UK become a light-touch AI
regulation” law haven” for Chat GPT!)
Private ordering – control?
• Contracts and licenses
• Eg Responsible AI Licensing
(RAIL)(FAccT ‘22)
• Model cards, model sheets
• Privacy policies
• EPSRC Generative AI terms of Service
project Jan-March 2023, report end
April !
• Technical controls
• Filters : eg Open AI NSFW filter
• API control eg Project December app
• Internal human moderation when
finetuning the model pre release
• Watermarks for generated content eg PAI
Guidelines for Synthetic Media
• Enforcement?
• Privity of contract
AGI-nising choices: Nuclear arm or buggy software?
"Should we let machines flood our
information channels with
propaganda and
untruth? Should we automate
away all the jobs, including the
fulfilling ones? Should we develop
nonhuman minds that might
eventually outnumber, outsmart,
obsolete and replace
us? Should we risk loss of control
of our civilization?" (source;
emphasis in original)
Future of Life Institute Open Letter, March
22, 2023
Narayanan and Kapoor https://aisnakeoil.substack.com/p/a-
misleading-open-letter-about-sci?
CREDITS
Parrot images by James Stewart, Edinburgh University ,
@datacontroversies
“after” various unknown artists, using MidJourney, 2023.
Stable Diffusion images created by and © Lilian Edwards, 2023, made
with own photo.
Image of Emily Bender with parrot © New York Times, 2023
Image on slide 12 taken with thanks from Thomas Hoppner
“ChatGPT, Bard & Co.: an introduction to AI for competition and
regulatory lawyers”, 2023 at https://www.hausfeld.com/en-
gb/what-we-think/competition-bulletin/chatgpt-bard-co-an-
introduction-to-ai-for-competition-and-regulatory-lawyers/

More Related Content

What's hot

Generative AI Art - The Dark Side
Generative AI Art - The Dark SideGenerative AI Art - The Dark Side
Generative AI Art - The Dark Side
Abhinav Gupta
 
Automate your Job and Business with ChatGPT #3 - Fundamentals of LLM/GPT
Automate your Job and Business with ChatGPT #3 - Fundamentals of LLM/GPTAutomate your Job and Business with ChatGPT #3 - Fundamentals of LLM/GPT
Automate your Job and Business with ChatGPT #3 - Fundamentals of LLM/GPT
Anant Corporation
 

What's hot (20)

The Future is in Responsible Generative AI
The Future is in Responsible Generative AIThe Future is in Responsible Generative AI
The Future is in Responsible Generative AI
 
Generative AI and law.pptx
Generative AI and law.pptxGenerative AI and law.pptx
Generative AI and law.pptx
 
How ChatGPT and AI-assisted coding changes software engineering profoundly
How ChatGPT and AI-assisted coding changes software engineering profoundlyHow ChatGPT and AI-assisted coding changes software engineering profoundly
How ChatGPT and AI-assisted coding changes software engineering profoundly
 
ChatGPT, Foundation Models and Web3.pptx
ChatGPT, Foundation Models and Web3.pptxChatGPT, Foundation Models and Web3.pptx
ChatGPT, Foundation Models and Web3.pptx
 
Generative AI: Past, Present, and Future – A Practitioner's Perspective
Generative AI: Past, Present, and Future – A Practitioner's PerspectiveGenerative AI: Past, Present, and Future – A Practitioner's Perspective
Generative AI: Past, Present, and Future – A Practitioner's Perspective
 
Unlocking the Power of Generative AI Models and Systems such as GPT-4 and Cha...
Unlocking the Power of Generative AI Models and Systems such as GPT-4 and Cha...Unlocking the Power of Generative AI Models and Systems such as GPT-4 and Cha...
Unlocking the Power of Generative AI Models and Systems such as GPT-4 and Cha...
 
Using the power of Generative AI at scale
Using the power of Generative AI at scaleUsing the power of Generative AI at scale
Using the power of Generative AI at scale
 
The Rise of the LLMs - How I Learned to Stop Worrying & Love the GPT!
The Rise of the LLMs - How I Learned to Stop Worrying & Love the GPT!The Rise of the LLMs - How I Learned to Stop Worrying & Love the GPT!
The Rise of the LLMs - How I Learned to Stop Worrying & Love the GPT!
 
Future of AI - 2023 07 25.pptx
Future of AI - 2023 07 25.pptxFuture of AI - 2023 07 25.pptx
Future of AI - 2023 07 25.pptx
 
Implementing Ethics in AI
Implementing Ethics in AIImplementing Ethics in AI
Implementing Ethics in AI
 
A Tutorial to AI Ethics - Fairness, Bias & Perception
A Tutorial to AI Ethics - Fairness, Bias & Perception A Tutorial to AI Ethics - Fairness, Bias & Perception
A Tutorial to AI Ethics - Fairness, Bias & Perception
 
Generative AI Art - The Dark Side
Generative AI Art - The Dark SideGenerative AI Art - The Dark Side
Generative AI Art - The Dark Side
 
What Are the Problems Associated with ChatGPT?
What Are the Problems Associated with ChatGPT?What Are the Problems Associated with ChatGPT?
What Are the Problems Associated with ChatGPT?
 
Automate your Job and Business with ChatGPT #3 - Fundamentals of LLM/GPT
Automate your Job and Business with ChatGPT #3 - Fundamentals of LLM/GPTAutomate your Job and Business with ChatGPT #3 - Fundamentals of LLM/GPT
Automate your Job and Business with ChatGPT #3 - Fundamentals of LLM/GPT
 
Large Language Models - Chat AI.pdf
Large Language Models - Chat AI.pdfLarge Language Models - Chat AI.pdf
Large Language Models - Chat AI.pdf
 
Introduction to LLMs
Introduction to LLMsIntroduction to LLMs
Introduction to LLMs
 
Artificial Intelligence Bill of Rights: Impacts on AI Governance
Artificial Intelligence Bill of Rights: Impacts on AI GovernanceArtificial Intelligence Bill of Rights: Impacts on AI Governance
Artificial Intelligence Bill of Rights: Impacts on AI Governance
 
An Introduction to Generative AI - May 18, 2023
An Introduction  to Generative AI - May 18, 2023An Introduction  to Generative AI - May 18, 2023
An Introduction to Generative AI - May 18, 2023
 
ChatGPT OpenAI Primer for Business
ChatGPT OpenAI Primer for BusinessChatGPT OpenAI Primer for Business
ChatGPT OpenAI Primer for Business
 
Generative AI, WiDS 2023.pptx
Generative AI, WiDS 2023.pptxGenerative AI, WiDS 2023.pptx
Generative AI, WiDS 2023.pptx
 

Similar to How to regulate foundation models: can we do better than the EU AI Act?

Impact of Generative AI in Cybersecurity - How can ISO/IEC 27032 help?
Impact of Generative AI in Cybersecurity - How can ISO/IEC 27032 help?Impact of Generative AI in Cybersecurity - How can ISO/IEC 27032 help?
Impact of Generative AI in Cybersecurity - How can ISO/IEC 27032 help?
PECB
 
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...
Edge AI and Vision Alliance
 
icon-aiincs-obusolini201809131800-190310184140.pptx
icon-aiincs-obusolini201809131800-190310184140.pptxicon-aiincs-obusolini201809131800-190310184140.pptx
icon-aiincs-obusolini201809131800-190310184140.pptx
yugandharadahiphale2
 
icon-aiincs-obusolini201809131800-190310184140.pptx
icon-aiincs-obusolini201809131800-190310184140.pptxicon-aiincs-obusolini201809131800-190310184140.pptx
icon-aiincs-obusolini201809131800-190310184140.pptx
yugandharadahiphale2
 

Similar to How to regulate foundation models: can we do better than the EU AI Act? (20)

Impact of Generative AI in Cybersecurity - How can ISO/IEC 27032 help?
Impact of Generative AI in Cybersecurity - How can ISO/IEC 27032 help?Impact of Generative AI in Cybersecurity - How can ISO/IEC 27032 help?
Impact of Generative AI in Cybersecurity - How can ISO/IEC 27032 help?
 
Generative Artificial Intelligence and Data Privacy: A Primer
Generative Artificial Intelligence and Data Privacy: A Primer Generative Artificial Intelligence and Data Privacy: A Primer
Generative Artificial Intelligence and Data Privacy: A Primer
 
Silicon Halton Meetup 108 - Is Your AI Invention Protectable?
Silicon Halton Meetup 108 - Is Your AI Invention Protectable?Silicon Halton Meetup 108 - Is Your AI Invention Protectable?
Silicon Halton Meetup 108 - Is Your AI Invention Protectable?
 
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...
 
influence of AI in IS
influence of AI in ISinfluence of AI in IS
influence of AI in IS
 
AI and ML Series - Introduction to Generative AI and LLMs - Session 1
AI and ML Series - Introduction to Generative AI and LLMs - Session 1AI and ML Series - Introduction to Generative AI and LLMs - Session 1
AI and ML Series - Introduction to Generative AI and LLMs - Session 1
 
Artificial Intelligence and Cybersecurity
Artificial Intelligence and CybersecurityArtificial Intelligence and Cybersecurity
Artificial Intelligence and Cybersecurity
 
Technologies in Support of Big Data Ethics
Technologies in Support of Big Data EthicsTechnologies in Support of Big Data Ethics
Technologies in Support of Big Data Ethics
 
Internet of Things: Trends and challenges for future
Internet of Things: Trends and challenges for futureInternet of Things: Trends and challenges for future
Internet of Things: Trends and challenges for future
 
icon-aiincs-obusolini201809131800-190310184140.pptx
icon-aiincs-obusolini201809131800-190310184140.pptxicon-aiincs-obusolini201809131800-190310184140.pptx
icon-aiincs-obusolini201809131800-190310184140.pptx
 
icon-aiincs-obusolini201809131800-190310184140.pptx
icon-aiincs-obusolini201809131800-190310184140.pptxicon-aiincs-obusolini201809131800-190310184140.pptx
icon-aiincs-obusolini201809131800-190310184140.pptx
 
Tutorial helsinki 20180313 v1
Tutorial helsinki 20180313 v1Tutorial helsinki 20180313 v1
Tutorial helsinki 20180313 v1
 
Data collection and enhancement
Data collection and enhancementData collection and enhancement
Data collection and enhancement
 
AI Cybersecurity: Pros & Cons. AI is reshaping cybersecurity
AI Cybersecurity: Pros & Cons. AI is reshaping cybersecurityAI Cybersecurity: Pros & Cons. AI is reshaping cybersecurity
AI Cybersecurity: Pros & Cons. AI is reshaping cybersecurity
 
Japan 20200724 v13
Japan 20200724 v13Japan 20200724 v13
Japan 20200724 v13
 
Artificial Intelligence in testing - A STeP-IN Evening Talk Session Speech by...
Artificial Intelligence in testing - A STeP-IN Evening Talk Session Speech by...Artificial Intelligence in testing - A STeP-IN Evening Talk Session Speech by...
Artificial Intelligence in testing - A STeP-IN Evening Talk Session Speech by...
 
Ethics of Analytics and Machine Learning
Ethics of Analytics and Machine LearningEthics of Analytics and Machine Learning
Ethics of Analytics and Machine Learning
 
DevOps Support for an Ethical Software Development Life Cycle (SDLC)
DevOps Support for an Ethical Software Development Life Cycle (SDLC)DevOps Support for an Ethical Software Development Life Cycle (SDLC)
DevOps Support for an Ethical Software Development Life Cycle (SDLC)
 
Algorithm Marketplace and the new "Algorithm Economy"
Algorithm Marketplace and the new "Algorithm Economy"Algorithm Marketplace and the new "Algorithm Economy"
Algorithm Marketplace and the new "Algorithm Economy"
 
Lesson 1 intro to ai
Lesson 1   intro to aiLesson 1   intro to ai
Lesson 1 intro to ai
 

More from Lilian Edwards

The death of data protection sans obama
The death of data protection sans obamaThe death of data protection sans obama
The death of data protection sans obama
Lilian Edwards
 

More from Lilian Edwards (20)

What Do You Do with a Problem Like AI?
What Do You Do with a Problem Like AI?What Do You Do with a Problem Like AI?
What Do You Do with a Problem Like AI?
 
The GDPR, Brexit, the UK and adequacy
The GDPR, Brexit, the UK and adequacyThe GDPR, Brexit, the UK and adequacy
The GDPR, Brexit, the UK and adequacy
 
Slave to the Algorithm 2016
Slave to the Algorithm  2016 Slave to the Algorithm  2016
Slave to the Algorithm 2016
 
Cloud computing : legal , privacy and contract issues
Cloud computing : legal , privacy and contract issuesCloud computing : legal , privacy and contract issues
Cloud computing : legal , privacy and contract issues
 
Privacy, the Internet of Things and Smart Cities
Privacy, the Internet of Things and Smart Cities Privacy, the Internet of Things and Smart Cities
Privacy, the Internet of Things and Smart Cities
 
From Privacy Impact Assessment to Social Impact Assessment: Preserving TRrus...
From Privacy Impact Assessment to Social Impact Assessment: Preserving TRrus...From Privacy Impact Assessment to Social Impact Assessment: Preserving TRrus...
From Privacy Impact Assessment to Social Impact Assessment: Preserving TRrus...
 
UK copyright, online intermediaries and enforcement
UK copyright, online intermediaries and enforcementUK copyright, online intermediaries and enforcement
UK copyright, online intermediaries and enforcement
 
The GDPR for Techies
The GDPR for TechiesThe GDPR for Techies
The GDPR for Techies
 
the Death of Privacy in Three Acts
the Death of Privacy in Three Actsthe Death of Privacy in Three Acts
the Death of Privacy in Three Acts
 
Revenge porn: punish, remove, forget, forgive?
Revenge porn: punish, remove, forget, forgive? Revenge porn: punish, remove, forget, forgive?
Revenge porn: punish, remove, forget, forgive?
 
From piracy to “The Producers?
From piracy to “The Producers?From piracy to “The Producers?
From piracy to “The Producers?
 
The Death of Privacy in Three Acts
The Death of Privacy in Three ActsThe Death of Privacy in Three Acts
The Death of Privacy in Three Acts
 
Police surveillance of social media - do you have a reasonable expectation of...
Police surveillance of social media - do you have a reasonable expectation of...Police surveillance of social media - do you have a reasonable expectation of...
Police surveillance of social media - do you have a reasonable expectation of...
 
IT law : the middle kingdom between east and West
IT law : the middle kingdom between east and WestIT law : the middle kingdom between east and West
IT law : the middle kingdom between east and West
 
What do we do with aproblem like revenge porn ?
What do we do with  aproblem like  revenge porn ?What do we do with  aproblem like  revenge porn ?
What do we do with aproblem like revenge porn ?
 
Slave to the Algo-Rhythms?
Slave to the Algo-Rhythms?Slave to the Algo-Rhythms?
Slave to the Algo-Rhythms?
 
9worlds robots
9worlds robots9worlds robots
9worlds robots
 
The death of data protection
The death of data protection The death of data protection
The death of data protection
 
The death of data protection sans obama
The death of data protection sans obamaThe death of data protection sans obama
The death of data protection sans obama
 
Cdas 2012, lilian edwards and edina harbinja
Cdas 2012, lilian edwards and edina harbinjaCdas 2012, lilian edwards and edina harbinja
Cdas 2012, lilian edwards and edina harbinja
 

Recently uploaded

一比一原版(Columbia毕业证书)哥伦比亚大学毕业证原件一模一样
一比一原版(Columbia毕业证书)哥伦比亚大学毕业证原件一模一样一比一原版(Columbia毕业证书)哥伦比亚大学毕业证原件一模一样
一比一原版(Columbia毕业证书)哥伦比亚大学毕业证原件一模一样
doypbe
 
一比一原版(QUT毕业证书)昆士兰科技大学毕业证如何办理
一比一原版(QUT毕业证书)昆士兰科技大学毕业证如何办理一比一原版(QUT毕业证书)昆士兰科技大学毕业证如何办理
一比一原版(QUT毕业证书)昆士兰科技大学毕业证如何办理
Airst S
 
一比一原版(UNSW毕业证书)新南威尔士大学毕业证如何办理
一比一原版(UNSW毕业证书)新南威尔士大学毕业证如何办理一比一原版(UNSW毕业证书)新南威尔士大学毕业证如何办理
一比一原版(UNSW毕业证书)新南威尔士大学毕业证如何办理
ss
 
一比一原版(ASU毕业证书)亚利桑那州立大学毕业证成绩单原件一模一样
一比一原版(ASU毕业证书)亚利桑那州立大学毕业证成绩单原件一模一样一比一原版(ASU毕业证书)亚利桑那州立大学毕业证成绩单原件一模一样
一比一原版(ASU毕业证书)亚利桑那州立大学毕业证成绩单原件一模一样
mefyqyn
 
Jual obat aborsi Bandung ( 085657271886 ) Cytote pil telat bulan penggugur ka...
Jual obat aborsi Bandung ( 085657271886 ) Cytote pil telat bulan penggugur ka...Jual obat aborsi Bandung ( 085657271886 ) Cytote pil telat bulan penggugur ka...
Jual obat aborsi Bandung ( 085657271886 ) Cytote pil telat bulan penggugur ka...
ZurliaSoop
 
一比一原版(JCU毕业证书)詹姆斯库克大学毕业证如何办理
一比一原版(JCU毕业证书)詹姆斯库克大学毕业证如何办理一比一原版(JCU毕业证书)詹姆斯库克大学毕业证如何办理
一比一原版(JCU毕业证书)詹姆斯库克大学毕业证如何办理
Airst S
 
一比一原版曼彻斯特城市大学毕业证如何办理
一比一原版曼彻斯特城市大学毕业证如何办理一比一原版曼彻斯特城市大学毕业证如何办理
一比一原版曼彻斯特城市大学毕业证如何办理
Airst S
 
Article 12 of the Indian Constitution law
Article 12 of the Indian Constitution lawArticle 12 of the Indian Constitution law
Article 12 of the Indian Constitution law
yogita9398
 
一比一原版伦敦南岸大学毕业证如何办理
一比一原版伦敦南岸大学毕业证如何办理一比一原版伦敦南岸大学毕业证如何办理
一比一原版伦敦南岸大学毕业证如何办理
Airst S
 

Recently uploaded (20)

一比一原版(Columbia毕业证书)哥伦比亚大学毕业证原件一模一样
一比一原版(Columbia毕业证书)哥伦比亚大学毕业证原件一模一样一比一原版(Columbia毕业证书)哥伦比亚大学毕业证原件一模一样
一比一原版(Columbia毕业证书)哥伦比亚大学毕业证原件一模一样
 
一比一原版(QUT毕业证书)昆士兰科技大学毕业证如何办理
一比一原版(QUT毕业证书)昆士兰科技大学毕业证如何办理一比一原版(QUT毕业证书)昆士兰科技大学毕业证如何办理
一比一原版(QUT毕业证书)昆士兰科技大学毕业证如何办理
 
Jim Eiberger Rental Agreement Redacted Former Lease.docx
Jim Eiberger Rental Agreement Redacted Former Lease.docxJim Eiberger Rental Agreement Redacted Former Lease.docx
Jim Eiberger Rental Agreement Redacted Former Lease.docx
 
一比一原版(UNSW毕业证书)新南威尔士大学毕业证如何办理
一比一原版(UNSW毕业证书)新南威尔士大学毕业证如何办理一比一原版(UNSW毕业证书)新南威尔士大学毕业证如何办理
一比一原版(UNSW毕业证书)新南威尔士大学毕业证如何办理
 
一比一原版(ASU毕业证书)亚利桑那州立大学毕业证成绩单原件一模一样
一比一原版(ASU毕业证书)亚利桑那州立大学毕业证成绩单原件一模一样一比一原版(ASU毕业证书)亚利桑那州立大学毕业证成绩单原件一模一样
一比一原版(ASU毕业证书)亚利桑那州立大学毕业证成绩单原件一模一样
 
Dematerialisation of securities of private companies
Dematerialisation of securities of private companiesDematerialisation of securities of private companies
Dematerialisation of securities of private companies
 
Chambers Global Practice Guide - Canada M&A
Chambers Global Practice Guide - Canada M&AChambers Global Practice Guide - Canada M&A
Chambers Global Practice Guide - Canada M&A
 
CASE STYDY Lalman Shukla v Gauri Dutt BY MUKUL TYAGI.pptx
CASE STYDY Lalman Shukla v Gauri Dutt BY MUKUL TYAGI.pptxCASE STYDY Lalman Shukla v Gauri Dutt BY MUKUL TYAGI.pptx
CASE STYDY Lalman Shukla v Gauri Dutt BY MUKUL TYAGI.pptx
 
Jual obat aborsi Bandung ( 085657271886 ) Cytote pil telat bulan penggugur ka...
Jual obat aborsi Bandung ( 085657271886 ) Cytote pil telat bulan penggugur ka...Jual obat aborsi Bandung ( 085657271886 ) Cytote pil telat bulan penggugur ka...
Jual obat aborsi Bandung ( 085657271886 ) Cytote pil telat bulan penggugur ka...
 
Career As Legal Reporters for Law Students
Career As Legal Reporters for Law StudentsCareer As Legal Reporters for Law Students
Career As Legal Reporters for Law Students
 
5-6-24 David Kennedy Article Law 360.pdf
5-6-24 David Kennedy Article Law 360.pdf5-6-24 David Kennedy Article Law 360.pdf
5-6-24 David Kennedy Article Law 360.pdf
 
Mischief Rule of Interpretation of statutes
Mischief Rule of Interpretation of statutesMischief Rule of Interpretation of statutes
Mischief Rule of Interpretation of statutes
 
一比一原版(JCU毕业证书)詹姆斯库克大学毕业证如何办理
一比一原版(JCU毕业证书)詹姆斯库克大学毕业证如何办理一比一原版(JCU毕业证书)詹姆斯库克大学毕业证如何办理
一比一原版(JCU毕业证书)詹姆斯库克大学毕业证如何办理
 
Understanding the Role of Labor Unions and Collective Bargaining
Understanding the Role of Labor Unions and Collective BargainingUnderstanding the Role of Labor Unions and Collective Bargaining
Understanding the Role of Labor Unions and Collective Bargaining
 
posts-harmful-to-secular-structure-of-the-country-539103-1.pdf
posts-harmful-to-secular-structure-of-the-country-539103-1.pdfposts-harmful-to-secular-structure-of-the-country-539103-1.pdf
posts-harmful-to-secular-structure-of-the-country-539103-1.pdf
 
Cyber Laws : National and International Perspective.
Cyber Laws : National and International Perspective.Cyber Laws : National and International Perspective.
Cyber Laws : National and International Perspective.
 
一比一原版曼彻斯特城市大学毕业证如何办理
一比一原版曼彻斯特城市大学毕业证如何办理一比一原版曼彻斯特城市大学毕业证如何办理
一比一原版曼彻斯特城市大学毕业证如何办理
 
Article 12 of the Indian Constitution law
Article 12 of the Indian Constitution lawArticle 12 of the Indian Constitution law
Article 12 of the Indian Constitution law
 
OVERVIEW OF LABOUR LAWS with Case Studies- ppt.ppt
OVERVIEW OF LABOUR LAWS with Case Studies- ppt.pptOVERVIEW OF LABOUR LAWS with Case Studies- ppt.ppt
OVERVIEW OF LABOUR LAWS with Case Studies- ppt.ppt
 
一比一原版伦敦南岸大学毕业证如何办理
一比一原版伦敦南岸大学毕业证如何办理一比一原版伦敦南岸大学毕业证如何办理
一比一原版伦敦南岸大学毕业证如何办理
 

How to regulate foundation models: can we do better than the EU AI Act?

  • 1. How to regulate foundation models: can we do better than the EU AI Act? Lilian Edwards Professor of Law, Newcastle University @lilianedwards Lilian.edwards@ncl.ac.uk April 2023
  • 2. What are large or “foundation” models? • GPT-2/3/3.5/4 (Open AI/Microsoft)(prompt to text)(2019 on) • “Large Language Model” or LLM • ChatGPT • DALL-E 2 (text to images – Google) • Stable Diffusion (open source – text to image) • HarmonAI – makes AI generated music (Stability) • CoPilot (prompt generates computer code – GitHub/OpenAI) • Meta Make-me-A-Video (text to video - Meta) • ERNIE ( Baidu, China) (prompt to text)
  • 3. eg Stable Diffusion : img to img (open source third party code)
  • 6.
  • 7.
  • 9. Integration into search, Feb 2023 New York Times
  • 10.
  • 11. Important (for law) features of large or “foundation” models • Generative – create text, images etc rather than merely classifying or predicting (ML) • Trained on unprecedentedly large datasets • Often scraped from “public” Internet • Impossible to manually review legality, privacy or harm of every item in datasets • Computationally expensive and retraining slow -> • large tech co dominance • GPT-4 training cost >$100mn • environmentally worrying • Training sets allow the model to assess probability of next word, pixel etc – not direct copying • Models are general , can have multiple uses, eg to write a party invite, a racist attack or provide customer support within an automated hiring system • Generated content increasingly difficult to distinguish from human-created content (disinfo, deepfakes) • Outputs may be “hallucinations” Hoppner, 2023
  • 12. Issues with large models PHASE 1 STOCHASTIC PARROTS • Don’t actually understand , just “parrot” • Bias, discrimination, misrepresentation and stereotyping of groups; hate speech PHASE 2 WILL NO-ONE THINK OF THE ARTISTS? • Image and video deepfakes • Pastiche • Copyright PHASE 3 FAKE NEWS ON STEROIDS • Fake news and “hallucination” (text + images] • Education & plagiarism • Digital Services Act PHASE 4 – YOU HAVE ZERO PRIVACY, GET OVER IT • GDPR
  • 13. Solution 1 : the EU AI Act and “GPAI” AIA “risk based” approach • Unacceptable risk – ‘Complete’ prohibition, 4 examples – Article 5 • High-risk –Fixed categories of risky domains, based on intended use ; “essential requirements” including dataset quality, human oversight – • Limited risk – Transparency obligations for a few AI systems (chatbots, deepfakes, emotion ID, biometric categorisation) – Article 52 • Minimal risk – Codes of conduct – Article 69 Photo Source: European Commission, Digital Strategy Website https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai
  • 14. Annex II - Products • Machinery • Toys • Recreational craft and watercraft • Lifts • Equipment and protective systems intended for use in potentially explosive atmospheres • Radio equipment • Pressure equipment • Cableway installations • PPE • Medical devices • [...] High-Risk AI systems (Designation) - Annexes II&III • Annex III - Services • Biometric identification and categorisation of natural persons; • Management and operation of critical infrastructure; • Education and vocational training; • Employment, workers management and access to self-employment; • Access to and enjoyment of essential private services and public services and benefit; • Law enforcement; • Migration, asylum and border control management; • Administration of justice and democratic processes
  • 15. • Compliance with requirements (Art.8); • Risk management system (Art.9); • Data and data governance (Art.10); (“data quality”) • “Training, validation and testing data sets shall be relevant, representative, free of errors and complete. They shall have the appropriate statistical properties, including, where applicable, as regards the persons or groups of persons on which the high-risk AI system is intended to be used.” • Technical documentation (Art.11); • Record-keeping (Art.12); • Transparency and provision of information to users (Art.13); • Human oversight (Art.14); • “Human oversight shall aim at preventing or minimising the risks to health, safety or fundamental rights that may emerge when a high- risk AI system is used in accordance with its intended purpose or under conditions of reasonably foreseeable misuse” • Accuracy, robustness and cybersecurity (Art.15). High-Risk AI systems (Requirements) – Arts. 8 - 15
  • 16.
  • 17. Definition of GPAI in EU AIA LGAIMs = Large General AI Models – Hacker, Engel and Mauer “Regulating ChatGPT and other Large generative AI Models”, February 2023 Issues : over-inclusive (no emphasis generality in “ability, task or output”); important for classification -> Council position ‘general purpose AI system’ means an AI system that is trained on broad data at scale, is designed for generality of output, and can be adapted to a wide range of tasks EU Parliament, March 2023
  • 18. Developers v deployers • Developers? • Akin to manufacturers – control and have knowledge of the training sets, weights, algorithms, content moderation etc (esp if closed source) • Most high-risk obligations arise at development stage (training & human fine-tuning stages) • Have practical power and economic gains • But can they handle unforeseeable uses/risks? Or can only tech giants  competition issues? Open source providers? • Deployers? [NB “users” in AIA!] • Originally only duties on deployer if they make a “substantial modification” to the AI system, ie become new provider • Practicality ?: May be impossible for them to fix or even audit issues of data quality etc without access to upstream source code, training datasets etc (often secret/proprietary – eg as with GPT-3) • “AI as a service” API/cloud model will be prevalent till models smaller “Pick n Mix?”
  • 19. Solutions in AIA process? • 13 May 2022, French presidency added amendment excluding GPAIs from AIA • Back and forth.. • Council position, Dec 2022, arts 4a-4c • GPAI deemed high risk if they “may be used as high risk AI systems or as components of high risk AI systems” • Unless “explicitly excluded all high-risk uses”, but not if not in “good faith” • European Parliament • All generative AI models to go into high risk if • they generate text that might be mistaken for human • And all deepfakes and AV content showing something that “never happened” unless an “obvious artistic work” • & Providers owe cooperation & transparency obligations to downstream users • Commission to tweak the high-risk obligations by delegated Acts..
  • 20. Foundation models v GPAI?? 19/4/23 “Foundation models” • “an AI system model that is trained on broad data at scale, is designed for generality of output, and can be adapted to a wide range of distinctive tasks.” • Eg Chat GPT , Stable Diffusion • “Trained on data scraped from entire Internet”? Not just if labelled • Stricter obligations - High risk ++? • Adds sustainability obligations + independent expert ex ante oversight • Documented analysis and testing throughout lifecycle • Disclosures re copyright in training set; filters to avoid delivery unlawful content “GPAI” • “AI system that can be used in and adapted to a wide range of applications for which it was not intentionally and specifically designed.” • ?? E.g. “unlabelled data that need further training by the provider, such as algorithms developed to recognise skin cancer” • Laxer regime, since obligations fall on high risk providers who build on the data (?) Deployers & value chain • “non-binding standard contractual clauses that regulate rights and obligations consistent with each party’s level of control”
  • 21. • Discrimination/ equality law • Liability (product liability for AI) • Copyright • Content, hate speech, libel, fake news (DSA) • Privacy & data protection • Personality & image rights • Competition law • See Hoppner https://papers.ssrn.com/sol3/paper s.cfm?abstract_id=4371681 • Apart from last though, only AIA and DMA allows for structural ex ante regulation • Private ordering? Solution 2 : everything but the AI Act!
  • 22. 1. Copyright & AI generated art: another entire talk • WaPost : “He used AI to win a fine-arts competition. Was it cheating? • One judge said the striking piece evoked Renaissance art. But some critics compared it to ‘entering a marathon and driving a Lamborghini to the finish line.’” • Effect on original artists? A tool or a replacement??
  • 23. • Boris Eldagsen The Electrician • “winner” Sony World Photography Awards, • 17 April 2023
  • 24. AI generated art and copyright • Often copyright in the art used as training dataset inputs (eg Rutowski) – but they are not part of the model • No direct copying though sometimes a perfect copy might emerge (memorisation) • Is there actual copying of inputs? • Is there (US) fair use or (UK/DSA) research or TDM exception? • Who owns the outputs? • Again almost never direct copies • “After the style of”.. • Derivative works? • Some partial solutions : opt-out; haveIbeentrained ; license, royalties or benefit sharing • Litigation!
  • 25. AI art litigation (US) • Getty v Stability, based on copying of input works copyright Getty • Transformative fair use? • Anderson vs Stability & Deviant Art • Aims to acquire rights over all OUTPUTS as derivative works from artistic works in training set • Defendants case to dismiss - • NB Stability is open source, so you can analyse the underlying training sets (cf Open (sic) AI & GPT-4)
  • 26. If you thought artists were p*ssed off by genAI.. … try the music industry!
  • 27. Can GPAI be compliant with GDPR? • AIA high risk data quality and transparency requirements actually say nothing about privacy of users • Machine learning process long regarded as dubious but • no federal DP law in US • no quality of privacy, confidentiality to data made public in US • GPAIs use permissionless public data (eg Common Crawl) • But - Replika decision, Italy, 2/2/23 • Primarily about exposure of children to unsuitable sexualized material • Unable to make valid contracts
  • 28. Italy DPA Garantie vs ChatGPT, 31 March 2023 “ “
  • 29.
  • 31. What next? • “reports of a 400% surge in VPN downloads in Italy”  • Spain, France, pan-European investigation by EDPB • Canada investigation • Does GPAI fundamentally challenge the GDPR? • If so, which gives?? (and does it take ML with it?) • The end of the data Wild West? • But are the privacy regulators really the right ones to take generative AI on? • (and will the UK become a light-touch AI regulation” law haven” for Chat GPT!)
  • 32. Private ordering – control? • Contracts and licenses • Eg Responsible AI Licensing (RAIL)(FAccT ‘22) • Model cards, model sheets • Privacy policies • EPSRC Generative AI terms of Service project Jan-March 2023, report end April ! • Technical controls • Filters : eg Open AI NSFW filter • API control eg Project December app • Internal human moderation when finetuning the model pre release • Watermarks for generated content eg PAI Guidelines for Synthetic Media • Enforcement? • Privity of contract
  • 33. AGI-nising choices: Nuclear arm or buggy software? "Should we let machines flood our information channels with propaganda and untruth? Should we automate away all the jobs, including the fulfilling ones? Should we develop nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us? Should we risk loss of control of our civilization?" (source; emphasis in original) Future of Life Institute Open Letter, March 22, 2023 Narayanan and Kapoor https://aisnakeoil.substack.com/p/a- misleading-open-letter-about-sci?
  • 34. CREDITS Parrot images by James Stewart, Edinburgh University , @datacontroversies “after” various unknown artists, using MidJourney, 2023. Stable Diffusion images created by and © Lilian Edwards, 2023, made with own photo. Image of Emily Bender with parrot © New York Times, 2023 Image on slide 12 taken with thanks from Thomas Hoppner “ChatGPT, Bard & Co.: an introduction to AI for competition and regulatory lawyers”, 2023 at https://www.hausfeld.com/en- gb/what-we-think/competition-bulletin/chatgpt-bard-co-an- introduction-to-ai-for-competition-and-regulatory-lawyers/

Editor's Notes

  1. P Hacker “The propensity of ChatGPT particularly to hallucinate when it does not find readymade answers can be exploited to generate text devoid of any connection to reality, but written in the style of utter confidence”
  2. Clip art getty photos Book covers Early product concept, ads, design architecture mock ups Can add shadows, extend out paintings, produce programme credits, instant editorial cartooning
  3. Eg anthropomorphisation of CBT chatbots, postmortem avatars* “Socio-economic “ harms Energy – very costly to train giant models
  4. Bans from fan fora – lots of fantasy art but very derivative =- in its nature! But – effort? “He started with a simple mental image — “a woman in a Victorian frilly dress, wearing a space helmet” — and kept fine-tuning the prompts, “using tests to really make an epic scene, like out of a dream.” He said he spent 80 hours making more than 900 iterations of the art, adding words like “opulent” and “lavish” to fine tune its tone and feel. He declined to share the full series of words he used to create his art, saying it is his artistic product, and that he intends to publish it later. “If there’s one thing you can take ownership of, it’s your prompt,” he said.” A tool or a copy?
  5. Transparency is to downstream deployer not data subject; data quality is not about whether personal data that was permissioned
  6. Eg Project December postmortem avatar app – app withdrawn by GPT-3 API control as inappropriate, but not on application of end-user Later they stopped asking you to fill in a form re the probity of your applicatiuon and now its just money!