Google DeepMind accused of breaking AI safety pledge in UK; Gets open letter from 60-plus lawmakers, says "troubling breach of trust"

Over half a dozen UK lawmakers have accused Google DeepMind of violating international AI safety commitments by releasing its Gemini 2.5 Pro model in March 2025 without proper safety documentation, calling it a "troubling breach of trust with governments and the public."The cross-party group, including digital rights campaigner Baroness Beeban Kidron and former Defence Secretary Des Browne, published an open letter through activist group PauseAI UK, claiming Google failed to honor the Frontier AI Safety Commitments signed at an international summit in February 2024.The controversy centers on Google's decision to release Gemini 2.5 Pro—which the company said outperformed competitors by "meaningful margins"—without immediately publishing a detailed model card containing safety testing information. The document was only released 22 days after the model became publicly available, with comprehensive safety details arriving over a month later.According to the letter shared with TIME, Google's delayed disclosure "sets a dangerous precedent" for AI development. The lawmakers criticised what they termed a "minimal model card" that lacked substantive details about external evaluations and failed to confirm whether government agencies like the UK AI Security Institute participated in testing."If leading companies like Google treat these commitments as optional, we risk a dangerous race to deploy increasingly powerful AI without proper safeguards," Lord Browne of Ladyton stated in the letter.At the 2024 summit, Google and other AI companies pledged to "publicly report" their models' capabilities and risk assessments while disclosing external testing involvement. The commitment was designed to provide transparency as AI systems become more powerful and potentially dangerous.A Google DeepMind spokesperson told Fortune the company stands by its "transparent testing and reporting processes" and is fulfilling public commitments, including the Seoul Frontier AI Safety Commitments. The spokesperson confirmed that Gemini 2.5 Pro underwent "rigorous safety checks, including by UK AISI and other third-party testers."However, Google only shared the model with the UK AI Security Institute after its March 25 release, according to TIME reporting. The lawmakers are demanding Google establish clear timelines for safety reporting and provide full transparency about testing partnerships for future releases.The letter reflects growing concerns about voluntary AI safety commitments, with similar criticism directed at other companies including Meta's Llama 4 and OpenAI's recent model releases lacking immediate safety documentation.