Happy Thursday! Last week OpenAI, Anthropic, and Google dominated headlines as Altman and Musk backed calls to slow AI development, Google launched Gemini 3.8 Live with real-time reasoning, and OpenAI acquired a camera AI startup for $300M.

AI REGULATORY UPDATE

Altman, Musk Back Amodei's Call to Slow AI

Anthropic CEO Dario Amodei has called for AI developers to slow frontier model development, earning rare public support from rivals Sam Altman of OpenAI and Elon Musk of xAI. Amodei proposed a three-step plan involving independent evaluators, industry coordination, and international agreements, warning that unchecked AI development could unleash uncontrollable bots causing hundreds of billions in damage within months. Altman confirmed OpenAI has already held internal discussions on pacing development and pledged to welcome independent evaluators with employee-level access. OpenAI Chief Scientist Jakub Pachocki also recently proposed voluntary pauses pending concrete safety standards. However, no formal talks between companies have begun, and Amodei acknowledged government support will likely be needed to make any slowdown legally viable.

AI ROBOTICS ANNOUNCEMENT

Huang Takes Trump's Live Call Onstage at Summit

Nvidia CEO Jensen Huang put President Trump on speakerphone during the All-In Summit, letting the crowd hear the call live. It marks the second time Trump has called Huang during a work event. During the call, Trump addressed concerns about automation, suggesting robots won't be taking over the world anytime soon. The moment drew significant attention given the high-profile setting and the two figures involved.

AI REGULATORY UPDATE

Microsoft Publishes Humanist AI Code of Conduct

Microsoft is releasing a 37-page humanist AI code of conduct, asserting that people matter more than AI amid escalating safety concerns in the industry. The move follows Anthropic CEO Dario Amodei's weekend call for a coordinated slowdown of AI development, after researchers warned that AI progress could outpace humanity's ability to safely deploy complex systems and control AI agents.

AI SAFETY GOVERNANCE

AI Labs Need Basic Security, Not Just Auditors

After an Anthropic researcher resigned over extinction fears, CEO Dario Amodei called for third-party audits of AI safety practices, winning support from OpenAI and Google executives. But cybersecurity experts say the real problem is more fundamental: AI labs are failing at basic network security. Experts point to poorly configured sandboxes, lack of real-time monitoring, and shared infrastructure as the core issues enabling AI agents to escape containment and access outside systems. OpenAI and Anthropic have begun improving observability of their models, but critics argue that fixing basic security hygiene would be more effective than outsourcing accountability to auditors.

GENERATIVE AI ENTERPRISE INTEGRATION

Claude Now Integrates Directly With Salesforce CRM

Anthropic has launched a Salesforce plugin for Claude in beta, giving sales teams AI-powered access to their accounts, opportunities, and pipeline data. The tool includes 37 skills covering daily sales tasks like call prep, deal reviews, pipeline forecasting, and CRM updates, with Claude acting only after seller approval. The plugin connects to both Salesforce and Slack, letting Claude generate morning briefings, prep meeting notes, score deals, and auto-update records after calls. Early adopters include GitLab, Siemens, and Legora, with 7,000 Salesforce sellers already using it. Available now in beta on all paid Claude plans.

NEW GENERATIVE AI LAUNCH

Google Launches Gemini 3.8 Live With Real-Time Reasoning

Google has released Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, its most advanced voice AI models to date. The models feature near-real-time reasoning, background tool execution, and support for 97 languages with automatic language detection and mid-conversation switching. Gemini 3.8 Live Extended Thinking scored 82.6 on the Artificial Analysis Speech to Speech Quality Index, topping rivals including GPT-Live-1-Astra. Both models are available via the Gemini API and Google AI Studio, with pricing starting at $0.005 per minute for audio input. Generated audio carries an invisible SynthID watermark to help detect misinformation.

AI ACQUISITION

OpenAI Acquires Camera AI Startup for $300M

OpenAI has acquired Glass Imaging, a smartphone camera AI company, for over $300 million, according to The Wall Street Journal. The Los Altos-based startup was founded by two former Apple engineers who led the development of Portrait Mode. Glass Imaging uses neural networks to improve smartphone photos in real time rather than through post-processing edits. The acquisition aligns with OpenAI's broader hardware ambitions, which include rumored smartphones, earbuds, and AI companion devices, as well as its $6.5 billion purchase of designer Jony Ive's startup io.

GENERATIVE AI LAUNCH

Claude Launches AI Suite for Financial Advisors

Anthropic has released Claude for Financial Advisors, a suite of connectors and workflow tools designed to reduce time spent on research, prep, and documentation. The product integrates with major industry platforms including Charles Schwab, BlackRock, Orion, Envestnet, and others, allowing advisors to access client data in one place. The release addresses a core industry problem: advisors spend only a sixth of their time in client meetings, with the rest consumed by administrative work. New skills cover meeting prep, portfolio analysis, compliance checks, and client onboarding, with firms able to customize workflows to their own service models.

GENERATIVE AI SAFETY POLICY

AI Safety Evaluators: Independent Watchdogs or Hired Hands?

Anthropic CEO Dario Amodei proposed embedding third-party safety evaluators inside frontier AI companies, granting them power to report incidents and publish findings without editorial control. OpenAI's Sam Altman also backed the idea, signaling a potential industry shift. Neither company has clarified key details about access, timing, or disclosure. Evaluators welcomed the proposal but warned it will only work if AI companies genuinely relinquish control. Past efforts have been hampered by restrictive NDAs, limited access, and tight timelines. Experts stress that evaluators need access to training data and checkpoints, not just finished models, to catch safety issues that models may be hiding during standard testing.

AI SAFETY STARTUP LAUNCH

New Startup Offers AI Agent Safety Certification

Former Anthropic employee Rune Kvist and ex-METR COO Rajiv Dattani have launched AIUC, a startup offering third-party auditing and certification for AI agents. The company has raised $55 million total, including a $40 million Series A led by Ribbit Capital, with customers including Cursor, Harvey, and ElevenLabs. AIUC developed a standard called AIUC-1, modeled on the cybersecurity framework SOC 2, and runs agents through 5,000 tests covering jailbreaks, hallucinations, and data leaks. The goal is to give enterprises independent verification of AI agent safety before deployment.

Upcoming Data & AI Conferences

Claude Founder House London 2026

Wed, Sep 23, 2026 | London, United Kingdom

Curated with AI by EMC2 AI

Brought to you by Estevan McCalley

Share this issue

Follow Estevan

Like this? Subscribe for the weekly AI roundup.