AI News for May 3, 2026 — 7min.ai

[OPINION] Jensen Huang slams AI doomer CEOs for adopting a 'God complex': Nvidia CEO Jensen Huang publicly rejected the wave of CEO predictions that AI will eliminate large swaths of the workforce, calling claims of existential risk and 50% job losses "ridiculous." Without naming names, he targeted Anthropic's Dario Amodei, who said AI could wipe out half of entry-level white-collar jobs, and Elon Musk, who claimed a 20% chance of human annihilation.
[RESEARCH] ARC-AGI-3 analysis pinpoints three systematic flaws in frontier reasoning: The ARC Prize Foundation analyzed 160 game runs from OpenAI's GPT-5.5 and Anthropic's Opus 4.7 on its interactive ARC-AGI-3 benchmark.
[NEWS] Chinese court bans firms from laying off workers to replace them with AI: A Chinese court ruled that companies cannot terminate employees solely to swap them for AI systems, the first major judicial check on AI-driven layoffs in the country.
[RESEARCH] Kimi K2.6 beats Claude, GPT-5.5, and Gemini in live coding contest: Moonshot AI's open-weights Kimi K2.6 won Day 12 of the AI Coding Contest's Word Gem Puzzle outright with 22 match points and a 7-1-0 record, ahead of Xiaomi's MiMo V2-Pro in second and OpenAI's GPT-5.5 in third.
[NEWS] OpenAI plans GPT-5.5 launch on May 5 with invite-only celebration: Sam Altman announced a small May 5 launch event for GPT-5.5 and shared an RSVP form, with OpenAI's Codex coding agent helping pick attendees from the replies.
[NEWS] Musk: 'I was a fool' giving OpenAI $38M in free funding that became an $800B company: Testifying over seven hours across three days in Oakland federal court, Musk told the jury he "was a fool" for providing $38 million in essentially free funding to a startup that became an $800 billion company, and argued OpenAI was "specifically meant to be for a charity that does not benefit any individual person." He is asking the court to unwind the nonprofit-to-for-profit conversion and remove Sam Altman and Greg Brockman from leadership.
[NEWS] ChatGPT turns on ad-tracking cookies by default for free users: OpenAI flipped marketing cookies to on by default for free ChatGPT users, sharing limited data including cookie IDs and email addresses with advertising partners to retarget users on Instagram and other third-party platforms.
[TOOLS] xAI launches Custom Voices: clone any voice from one minute of audio: xAI rolled out "Custom Voices," a feature that turns roughly a minute of natural speech recorded in the xAI console into a deployable voice clone in under two minutes.
[RESEARCH] Philosophy Bench: Claude refuses tasks, Grok complies, Gemini bends easily: A new Philosophy Bench by Benedict Brady runs frontier models through 100 ethically loaded everyday scenarios, from a VP of Sales pressuring for confidential customer data to a doctor bypassing protocol to enroll a minor in an oncology trial.
[NEWS] Academy bars AI-generated actors and screenplays from Oscars: The Academy of Motion Picture Arts and Sciences ruled that only performances "credited in the film's legal billing and demonstrably performed by humans with their consent" qualify for acting awards.

About 7min.ai · Subscribe to newsletter