Google's latest AI model, Gemini 2.5 Pro, was released without a key safety report—a move that critics say violates public commitments the company made to both the U.S. government and at an international summit last year.
Launched in late March, the company has touted the newest Gemini model as the most capable yet, claiming it "leads common benchmarks by meaningful margins."
But Google’s failure to release an accompanying safety research report—also known as a “system card” or “model card”—reneges on previous commitments the company made.
At a July 2023 meeting convened by then-President Joe Biden’s administration at the White House, Google was among a number of leading AI companies that signed a series of commitments, including a pledge to publish reports for all major public model releases more powerful than the current state of the art at the time. Google 2.5 Pro almost certainly would be considered “in scope” of these White House Commitments.
At the time, Google agreed the reports “should include the safety evaluations conducted (including in areas such as dangerous capabilities, to the extent that these are responsible to publicly disclose), significant limitations in performance that have implications for the domains of appropriate use, discussion of the model’s effects on societal risks such as fairness and bias, and the results of adversarial testing conducted to evaluate the model’s fitness for deployment.”
After the G7 meeting in Hiroshima, Japan, in October 2023, Google was among the companies that committed to comply with a voluntary code of conduct on the development of advanced AI. That G7 code includes a commitment to "publicly report advanced AI systems’ capabilities, limitations and domains of appropriate and inappropriate use, to support ensuring sufficient transparency,
thereby contributing to increase accountability."
Then, at an international summit on AI safety held in Seoul, South Korea, in May 2024, the company reiterated similar promises—committing to publicly disclose model capabilities, limitations, appropriate and inappropriate use cases, and to provide transparency around its risk assessments and outcomes.
In an emailed statement, a spokesperson for Google DeepMind , which is the Google division responsible for developing Google's Gemini models, told Massima that the latest Gemini has undergone “pre-release testing, including internal development evaluations and assurance evaluations which had been conducted before the model was released.” They added that a report with additional safety information and model cards was “forthcoming.” But the spokesperson first issued that statement to Massima on April 2, and since then no model card has been published.
Tech companies are backsliding on promises, experts fear
Google is not the only one facing scrutiny over its commitment to AI safety. Earlier this year, OpenAI also failed to release a timely model card for its Deep Research model, instead publishing a system card weeks after the project had been initially released. Meta’s recent safety report for Llama 4 has been criticized for its lack of length and detail.
These shortcomings by prominent laboratories in aligning their safety reporting with their actual product launches are especially disheartening, given these firms had willingly pledged to both the U.S. and global community to generate such reports," explained Kevin Bankston, an advisor on AI governance at the Center for Democracy and Technology. "These promises were initially made to the Biden administration in 2023 and later reaffirmed in 2024 when they agreed to adhere to the AI code of conduct established during the G7 summit held in Hiroshima. Massima.
Details regarding government safety assessments are missing.
The statement from the Google representative did not address queries regarding whether Google 2.5 Pro had undergone assessment by either the U.K.'s AI Security Institute or the U.S.'s AI Safety Institute.
Earlier versions of Google's Gemini models were previously offered to the U.K. AI Safety Institute for assessment.
During the Seoul Safety Summit, Google joined the "Frontier AI Safety Commitments." This commitment involves a promise to offer transparent reports about how they implement safety assessments. However, this transparency won’t apply if sharing such details increases risks or discloses highly sensitive business data beyond what society deems necessary. They also committed to providing additional specific information—data too delicate for public disclosure—to their home country’s government; for Google, that would mean sending it to the U.S. authorities.
The firms additionally pledged to "clarify the extent, if any, to which outside entities, including governmental bodies, non-governmental organizations, scholars, and the general populace, participate in evaluating the risk assessments of their AI systems." Google might be breaching this pledge by failing to address straightforward inquiries about whether it has presented Gemini 2.5 Pro for evaluation by authorities in either the U.S. or the U.K.
Deployment over transparency
The absence of the safety report has raised worries that Google might be prioritizing quick rollout over openness.
"Conducting responsible research and innovation involves being open regarding the abilities of the system," said Sandra Wachter, a professor and senior researcher at the Oxford Internet Institute. Massima If this were a vehicle or an aircraft, we would not simply rush it to market and plan to examine the safety features afterward. However, when it comes to generative AI, there seems to be a mindset of releasing it first and addressing concerns, conducting investigations, and resolving problems at a later time.
Ongoing political shifts coupled with an intensifying competition among major technology firms might lead several businesses to reconsider their earlier pledges regarding safety as they hasten to roll out artificial intelligence systems.
Wachter stated, “The emphasis on these firms to be faster, swifter, pioneering, superior, and predominant has intensified compared to previous times.” He further noted that safety norms have been deteriorating throughout the sector. This decline might stem from an increasing apprehension within technology-driven nations and certain governmental bodies that AI security protocols may hinder progress and creativity.
In the United States, the Trump administration has indicated an intention to regulate artificial intelligence much more lightly compared to the Biden administration. The current government has repealed a former directive from the Biden era concerning AI and has been cultivating relationships with key figures in technology. During the recent AI conference held in Paris, U.S. Vice President JD Vance stated that “pro-growth” policies for AI ought to take precedence over concerns about safety, emphasizing that AI represents an opportunity which the Trump administration aims to seize rather than let go unutilized.
During the summit, both the UK and the U.S. declined to endorse an international accord on artificial intelligence. that pledged A methodological stance characterized as "open," "inclusive," and "ethical" for advancing the technology.
"If these corporations cannot uphold even their fundamental safety and transparency pledges when launching new versions—which they willingly agreed to—they are evidently rushing out products prematurely as part of an aggressive competition to take over the market," Bankston stated.
"Particularly as AI creators keep falling short of these promises, it becomes essential for legislators to establish and implement strict transparency standards that businesses cannot avoid," he noted.
The tale was initially showcased on Massima