Grant Harvey

OpenAI shipped GPT-5.2, launched a ChatGPT app ecosystem, pushed AI safety research, and hinted at what’s next.

Google and Microsoft are taking radically different paths to AGI, as ex-DeepMind co-founders Demis Hassabis and Mustafa Suleyman reveal.

OpenAI launches GPT-5.2 with a 400K context window, stronger reasoning, and enterprise-ready coding features designed to handle full codebases in a single request.

Most companies use AI far below its potential. OpenAI’s new enterprise report reveals huge productivity gains and how frontier firms are pulling ahead fast.

Amazon unveiled 50+ AI updates at re:Invent, including new Nova models, autonomous agents, Bedrock upgrades, and cutting-edge Trainium hardware.

Remember when everyone said AI would only replace tech jobs? Turns out, they were looking at the tip of the iceberg. MIT and Oak Ridge National Laboratory just released Project Iceberg, a massive simulation that tracked 151 million US workers across 32,000 skills and 923 occupations to figure out which jobs AI can already automate […]

Anthropic found that AI models trained with reward-hacking shortcuts can develop deceptive, sabotaging behaviors.

Microsoft’s new Agent 365 gives IT full visibility and control over every AI agent in the organization, solving security, governance, and shadow-agent risks.

Anthropic releases a transparent political-bias test across major AI models, revealing how Claude, GPT-5, Gemini, Grok, and Llama handle even-handed responses.

Satya Nadella faces tough questions from Dwarkesh and Dylan Patel in a rare deep-dive interview on Microsoft’s AI strategy, data centers, and agents.

Researchers from Carnegie Mellon and Stanford compared AI agents to human workers, revealing key trade-offs in speed, cost, and quality.

AI agents aren’t replacing freelancers anytime soon. Scale AI’s benchmark shows they complete just 2–3% of real tasks—revealing limits of automation hype.