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: Responsible AI guidelines for startups, enforcing EU rules, new frontier model features, economic cases for generative AI, interviews with Anthropic’s Dario Amodei and OpenAI’s Mira Murati, and more.

Policy

Americas:

• Responsible Innovation Labs convenes cross-sector coalition of VCs and experts including the U.S. Dept. of Commerce to develop responsible AI guidelines for startups
• Everything you need to know about the U.S. government’s efforts to regulate AI (Fast Company)
• Senators Thune and Klobuchar team up on ‘light-touch’ AI bill
• California gov. vetoes bill to ban driverless AV trucks (more here)

Europe:

• EU warns TikTok and Meta over Slovakia election lies under the Digital Services Act
• UK dials up fight with Meta over encryption in line with the new Online Safety Bill (more here)
• EU tells Apple chief to 'open up' to rivals under the Digital Markets Act

More:

• UN deadlocked over regulating AI and here's why it will be hard
• How China’s new AI rules could affect U.S. companies
• UAE wants to disrupt the AI superpowers

Responsible AI

Safety and alignment:

• Improving AI with Anthropic’s CEO Dario Amodei (more here)
• Where we go from here with OpenAI's Mira Murati, CTO (more here)

Information:

• “Reversal Curse” paper asks can LLMs trained on A is B infer automatically that B is A?
• Why Google’s VP Elizabeth Reid isn’t rushing AI search
• FCC Chair Rosenworcel proposes to restore net neutrality rules (more here)
• ICYMI: An AI content deluge is coming. Grammarly’s CEO Rahul Roy-Chowdhury wants to help

Sciences and health:

• Finding battery minerals with AI, with KoBold Metals’ co-founder Josh Goldman
• Programmable DNA machines offer general-purpose computing
• GE HealthCare awarded a $44M grant to develop AI ultrasound technology, with focus on low-and-middle income countries

Identity, security, and data:

• Metaphysic’s CEO Tom Graham wants to arm us in the fight against deepfakes
• DHS announces new facial recognition and analysis policies and proposes one common platform for reporting cyber incidents
• Togo harnesses AI to deliver responsive services in data scarce environment

Resources:

• LLMs in Production: A virtual conference (MLOps Community)
• Age Assurance: Guiding principles and best practices (Digital Trust & Safety Partnership)
• The Responsible AI Leadership Summit (Credo.ai)
• Compute and AI: Guide and policy considerations (AI Now Institute)

Where AI is going

Capabilities:

• ChatGPT can now search the web in real time and see, hear, and speak
• Spotify can translate podcasts into new languages
• Meta’s AI chatbot plan includes a ‘sassy robot’ for younger users

Economics and competition:

• Explosive growth from AI automation: a review of the arguments
• A16z on the economic case for generative AI and Sequoia on its ‘Act Two’
• The FTC just hit Amazon with a major antitrust lawsuit (more here)

Future of work:

• Hollywood writers' contract deal includes historic AI rules
• DOD Task Force Lima focuses on responsible use of generative AI in its department

Releases and raises:

• Releases: Baichuan’s series of multilingual LLMs, Generative AI by Getty Images, Bard’s fact check features, Cloudflare’s Workers AI, Windows’ AI updates
• Raises: Anthropic ($4B), Pryon (data, $100M), HiddenLayer (security, $50M), Secoda (data, $14M); OpenAI allegedly nears an $80-90B valuation (more here)

Partnerships and splits:

• Cisco makes largest ever acquisition, buying Splunk for $28B
• How Microsoft CTO Kevin Scott helped forge the OpenAI deal
• GGV splits off China business following congressional panel probe

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