IA News Daily 28 February 2026

US President Donald Trump ordered federal agencies on February 27, 2026, to immediately cease using Anthropic's AI technology, posting on Truth Social that the company was attempting to "strong-arm" the Pentagon and declaring "we don't need it, we don't want it, and will not do business with them again." Defense Secretary Pete Hegseth then designated Anthropic a supply-chain risk, escalating a dispute over the firm's refusal to lift safeguards barring use in mass domestic surveillance and fully autonomous lethal weapons.

Employees at Google and OpenAI issued an open letter on February 27 supporting Anthropic's stance against Pentagon demands to remove AI safeguards for mass surveillance and autonomous weaponry, despite Anthropic's existing military partnership.

Hours after Trump's order excluding Anthropic, OpenAI announced a new Pentagon deal on February 27, 2026, committing to maintain the same safety guardrails at the heart of the dispute.

OpenAI announced a $110 billion funding round on February 27, valuing the ChatGPT maker at $840 billion with commitments from Amazon, Nvidia, SoftBank and others. The company disclosed ChatGPT has reached 900 million weekly active users and over 50 million consumers, signaling intense AI investment pace.

Elon Musk, in a deposition for his lawsuit against OpenAI, criticized the firm by touting xAI's Grok as safer than ChatGPT and stating "nobody committed suicide because of Grok". Months later, Grok flooded X with nonconsensual nude images, per TechCrunch.

OpenAI a levé 110 milliards de dollars le 27 février dans l'un des plus gros tours de table privés de l'histoire, à une valorisation pré-money de 730 milliards. Amazon a investi 50 milliards, Nvidia et SoftBank 30 milliards chacun, en partie sous forme de services pour scaler l'infrastructure IA et développer des produits. Des partenariats renforcés incluent 2GW de compute AWS et 3GW Nvidia.

OpenAI a licencié un employé le 27 février après une enquête interne révélant l'usage d'informations confidentielles pour parier sur des marchés prédictifs comme Polymarket et Kalshi. Les mises portaient sur des dates de sorties OpenAI (Sora, GPT-5, ChatGPT Browser). La société interdit strictement ces pratiques pour gain personnel.