May 24, 2024

This Week in Responsible AI, May 24, 2024

This Week in Responsible AI, May 24, 2024

I was on vacation for a couple of weeks, so this issue is a jumbo one!

Labor

• Tech workers speak up about rushing "AI" to market

• “I hope this isn’t for weapons.” How Syrian data workers train AI

Tools and Frameworks

• Measuring User Experience Inclusivity in Human-AI Interaction via Five User Problem-Solving Styles

• Fishing for Magikarp: Automatically Detecting Under-trained Tokens in Large Language Models

• Operationalizing content moderation "accuracy" in the Digital Services Act

• Building, Shifting, & Employing Power: A Taxonomy of Responses From Below to Algorithmic Harm

• Algorithmic Harms and Algorithmic Wrongs

• Energy Star Ratings for AI Models

• Societal Adaptation to Advanced AI

• Towards Guaranteed Safe AI: A Framework for Ensuring Robust and Reliable AI Systems

• Introducing the Frontier Safety Framework

• AI Safety Institute releases new AI safety evaluations platform

Fakes

• ‘I Want That Sweet Baby’: AI-Generated Kids Draw Predators On TikTok And Instagram

• White House pushes tech industry to shut down market for sexually abusive AI deepfakes

• We have to stop ignoring AI’s hallucination problem

• Meet AdVon, the AI-Powered Content Monster Infecting the Media Industry

• As AI becomes more human-like, experts warn users must think more critically about its responses

• Google promised a better search experience — now it’s telling us to put glue on our pizza

• Mastermind behind Biden AI robocalls faces potential $6M fine from FCC

Copyright

• ChatGPT to lose voice over Johansson similarity

• Consent and Compensation: Resolving Generative AI’s Copyright Crisis

• How African NLP Experts Are Navigating the Challenges of Copyright, Innovation, and Access

Bias

• Humane Intelligence Algorithmic Bias Bounty

• The Dark Side of Dataset Scaling: Evaluating Racial Classification in Multimodal Models

• Perceptions in Pixels: Analyzing Perceived Gender and Skin Tone in Real-world Image Search Results

• The Silicon Ceiling: Auditing GPT's Race and Gender Biases in Hiring

Transparency

• The CLeAR Documentation Framework for AI Transparency: Recommendations for Practitioners & Context for Policymakers

• Foundation Model Transparency Reports

• TikTok will automatically label AI-generated content created on platforms like DALL·E 3

• Transparency in the Wild: Navigating Transparency in a Deployed AI System to Broaden Need-Finding Approaches

• Using Degeneracy in the Loss Landscape for Mechanistic Interpretability

• The Local Interaction Basis: Identifying Computationally-Relevant and Sparsely Interacting Features in Neural Networks

Law/Policy

• US intelligence agencies’ embrace of generative AI is at once wary and urgent

• EU member states sign off on the AI Act

• Colorado governor signs sweeping AI regulation bill

• UK watchdog looking into Microsoft AI taking screenshots

• Data and Society policy brief: AI governance needs sociotechnical expertise

• Writers and publishers in Singapore reject a government plan to train AI on their work

• Defining AI incidents and related terms

• The FTC has some research questions about AI that they would love help on

• Computational Law and Access to Justice

• The algorithmic administration: automated decision-making in the public sector

Other

• Promises and pitfalls of artificial intelligence for legal applications

• AI Chatbot Credited With Preventing Suicide. Should It Be?

• How news coverage, often uncritical, helps build up the AI hype

• Generative AI red teaming challenge: transparency report

• OpenAI disbands its 'safety' team

• Meta’s new AI council is composed entirely of white men

• Google’s call-scanning AI could dial up censorship by default, privacy experts warn

• LLMs’ Data-Control Path Insecurity

Compiled by Leif Hancox-Li