AI News Roundup — March 7, 2026
OpenAI's robotics chief quits over the Pentagon deal, DOGE deploys ChatGPT to gut the National Endowment for the Humanities, and the AI-jobs debate becomes a CEO cage match as Benioff pushes back against Dorsey's layoff predictions.
OpenAI Robotics Chief Caitlin Kalinowski Quits Over Pentagon Deal
â Caitlin Kalinowski, head of OpenAI's robotics division and a former Meta hardware exec, resigned in protest of OpenAI's Pentagon agreement. She's the highest-profile departure yet over the defense deal â and it comes just days after OpenAI said it was revising the arrangement. The exit raises questions about internal dissent at OpenAI as its military ambitions collide with employee values.
DOGE Deploys ChatGPT to Gut the National Endowment for the Humanities
â Elon Musk's DOGE agency has begun using ChatGPT inside the National Endowment for the Humanities to review and recommend cuts to arts and humanities funding. The move â using a commercial AI tool to make federal funding decisions â is drawing fierce criticism from academics, artists, and civil liberties groups who say automated systems shouldn't decide which cultural programs survive.
Google Gives Sundar Pichai a $692M Pay Package
â Alphabet approved a three-year compensation package for CEO Sundar Pichai worth approximately $692 million â reflecting Google's AI-driven revenue surge and the existential importance of the Gemini bet. It's one of the largest CEO pay packages in tech history.
Salesforce CEO Benioff Pushes Back on AI Layoff Hysteria
â Marc Benioff publicly disputed the wave of AI-driven layoff predictions, arguing the narrative is overblown and that AI is creating more roles than it's eliminating at Salesforce. His comments directly contrast with Jack Dorsey's prediction that "most companies" would follow Block's lead in halving headcount.
TechCrunch: "A Roadmap for AI, If Anyone Will Listen"
â Connie Loizos examines how the Anthropic-Pentagon rupture has exposed the glaring absence of coherent AI regulation. The piece argues the US is flying blind â no clear rules for military AI use, no framework for supply-chain designations, and no consensus on how safety-focused labs should interact with defense.
Grammarly's "Expert Review" Feature Has No Actual Experts
â TechCrunch and Wired investigations revealed that Grammarly's new "Expert Review" feature uses expert identities without their explicit consent and doesn't actually involve the named experts reviewing user text. A cautionary tale of AI product marketing outrunning reality.
OpenClaw Community Grows as Open-Source AI Alternative
â The OpenClaw open-source AI project held a packed Manhattan meetup, with supporters championing it as a "vital counterpoint to Big AI." The grassroots movement reflects growing demand for transparent, community-governed AI systems. AI Chief","published_at","2026-03-08T16:00:00.000Z","created_at","2026-03-08 16:01:21","actionData","errors"]