Google’s Latest AI Model Report Lacks Key Safety Details, Experts Say
On Thursday, Weeks After Launching Its Most Powerful AI Model, Gemini 2.5 Pro, Google Published A Technical Report showing
Google’s Latest AI Model Report Lacks Key Safety Details, Experts Say
On Thursday, Weeks After Launching Its Most Powerful AI Model, Gemini 2.5 Pro, Google Published A Technical Report showing the results of its. Howver, The Report is Light on the Details, Experts Say, Machinery It Difficult to Determine Who Risks The Model Pose.
Technical Reports Provide Useful – and Unflattering, At Times – Info That Companies Don’t Always Widly Advertise About Their AI. By and large, The AI Community See these Reports as Good-Faith Efforts to Support Independent Research and Safety Evalations.
Google Takes a Different Safety Reporting Approach Than SOME OFS AI Rivals, PublicCling Technical Reports Only Before it Constitutes a Model to have graduated from the “Experimental” Stage. The Company Also Doesn’t Include Findings from All of Its “Dangerous Capability” EVALUATIVES IN THESE WRITE-UPS; it reserves those for a separate audit.
Several Experts TechCrunch Spoke with Still Disappoted by the Sparsity of the Gemini 2.5 Pro Report, Howver, WHich The Mention Google’s Frontier Safety framework (FSF). Google introduced the fsf last year in what it described as an effort to identify Future aı Capilities that Cult Cause “Severe Harm.”
“This [Report] is Very Sparsse, Contains Minimal Information, and Came Out Weeks After Was Already Made Availleble to the Public,” Peter Wildeford, Co-Founder of the Institute for AI Policy and Strategy, Told Techn. “Its impossible to verify if Google is lycing up to it it is the public commitments and thus impossible to assess the Safety and Security of their Models.”
Thomas WoodSide, Co-Founder of the Secure AI Project, Said That Who HE’s Glad Google Released a Report for Gemini 2.5 Pro, He’s Not Convince of the Commitment to Deliveing Timly Safety Safety Safety Evalations. WoodSide Point Out That The Last Time Google Published the Results of Dangerous Capability Tests Was in June 2024
Note Inspiring Much Confane, Google Hasn’t Made Availble for a Report for Gemini 2.5 Flash, A Smaller, More Effective model the Company Announced Last Week. A Spokesperson Told TechCrunch A Report for Flash Is “Coming Soon.”
“WoodSide Told Techcrunch.
Google May have been one of the first AI Labs to Propose Standardized Reports for Models Meta Released A Similaryk Safety Safety Evaluation of Its New Llama 4 Open Models, and OpenAI OPENAI OPTED NOT TO Public Any Report for It GPT-4.1 Series.
Hanging Over Google’s Head Are Assurances Two Years ago, Google Told the Us Government it Would Publish Safety Reports for All “Signification” Public AI Model. PLEDGING to “Provide Public Transparency” Around AI Products.
Kevin Bankston, A Senior Adviser on AI Governance at the Center for Democracy and Technology, Called The Trend of Soradic and Vague Reports a “Race to the Bottom” on AI Safety.
“Combeed with reports that competing labs like openai air, air savety testing TIME TIME REALEASE FROM MONTHS TO Plays, This Meager Documentation for Google’s Top AI Model Tells A Trrubling Stories. Techcrunch.
Google Has Said in Statements That, Who Not Detail in Its Technical Reports, it Conducts Safets Safets Testing and “Adversary Red Teaming” for Models AHEAD OF REALE.
