Welcome to this week’s edition of What's Happening in AI, covering the latest in policy, responsible AI, and where AI is going for the Responsible Innovation Labs community.

Below: EU opens its supercomputers to startups, how AI could reshape governments, and more. First, this week’s big story: The White House secures a second round of voluntary commitments (and news from today’s AI Insight Forum)

• Why this matters: In the context of broader regulation efforts, companies leading on AI (ex. Inflection, OpenAI, Microsoft) are working to deliver on these commitments
• What's next: Regulatory action from Congress (Schumer, Blumenthal and Hawley, AI transparency hearing), Executive Order (White House, California and other states, agencies, etc.)
• Who’s signed: July (Amazon, Anthropic, Google, Inflection, Meta, Microsoft, OpenAI), September (Adobe, Cohere, IBM, Nvidia, Palantir, Salesforce, Scale AI, Stability)
• Consider: Role of technology leaders to help shape effective regulation (ex. 20+ executives at AI Insight Forum, Microsoft’s Brad Smith in a Senate hearing this week)
• Comments: “Imperative”, “the right next step”, “trust and safety at the center of the most powerful AI systems”, “ongoing collaboration”

Policy

Europe:

• EU to let ‘responsible’ AI startups train models on its supercomputers
• UK’s Frontier AI Taskforce announces its advisory board
• French lawmaker challenges transatlantic data deal before EU court

Asia:

• Shenzhen to open specific sections of city highways for autonomous vehicle trials

Global:

• G7 countries commit to AI code of conduct
• How Data Protection Authorities are de facto regulating generative AI
• AI regulation around the world (note China’s most recent interim policy)
• Recommendation for participation of emerging economies in a global AI summit (US NAIAC)

Responsible AI

Information and data:

• Google to require disclosures of AI content in political ads
• Amazon to require some authors to disclose the use of AI material
• Microsoft’s new commitment and Adobe’s indemnity clause: designed to offer legal protection for AI copyright infringement challenges

Society and alignment:

• OpenAI CEO Sam Altman on ChatGPT, Future of AI, AGI, and more
• Google pledges $20M for responsible AI grants fund
• Brookings on potential challenges and proposals for US AI regulation
• ICYMI: Microsoft on India’s AI Opportunity

Privacy, safety, and security:

• Generative AI's biggest security flaw, prompt injection, is not easy to fix
• How China demands tech firms reveal hackable flaws in their products
• ICYMI: 54 Attorneys General call on Congress to study AI and its harmful effects on children

Climate:

• AI technology behind ChatGPT was built in Iowa — with a lot of water
• Apple unveils its first carbon neutral products

Resources:

• Two real-life examples of why limiting permissions works: Lessons from AWS’ Customer Incident Response Team (AWS)
• Digital threats from East Asia increase in breadth and effectiveness (Microsoft)
• How to foster belonging and inclusion, with Ellen Pao (Multitudes)
• Workshop on responsible and open foundation models (Princeton-Stanford)

Where AI is going

Ethics and humanity:

• How the AI revolution will reshape the world (Inflection AI CEO Mustafa Suleyman)
• AI and Leviathan: Part III (economist Samuel Hammond)
• What if AI treats humans the way we treat animals? (Vox)

Autonomous and synthetic:

• Learning to let go (of the wheel)
• Meet 'PIBOT,' the robot who can fly planes
• AI voices are taking over the internet

Compute:

• Cerebras CEO Andrew Feldman on the world's largest AI processor
• NVIDIA’s TensorRT-LLM supercharges LLM inference on NVIDIA H100 GPUs

Releases, raises, and partnerships:

• Releases: Claude Pro (Anthropic), Persimmons-8B (Adept), Intuit Assist, Roblox Assistant, Ant Group's Financial LLM
• Raises: Imbue ($200M), D-Matrix ($110M), AI Grant companies – batch 2
• Partnerships: India’s Reliance and Nvidia

Sign up for weekly updates in your inbox here. What's happening in AI is shaped by our Charter themes and Responsible AI work.