Google's recent publication of a technical report on its powerful AI model, Gemini 2.5 Pro, has raised eyebrows among experts, who claim the report is lacking in crucial details, making it difficult to assess the model's potential risks. The report's sparsity has sparked concerns over transparency and accountability in the AI industry, with some experts calling it a "race to the bottom" on AI safety.

The AI community generally views technical reports as good-faith efforts to support independent research and safety evaluations. However, Google's approach to safety reporting differs from its rivals, as it only publishes reports once a model has graduated from the "experimental" stage. Moreover, the company reserves findings from its "dangerous capability" evaluations for a separate audit, which has led to criticism over the lack of transparency.

Experts have expressed disappointment over the Gemini 2.5 Pro report, which fails to mention Google's Frontier Safety Framework (FSF), introduced last year to identify future AI capabilities that could cause "severe harm." Peter Wildeford, co-founder of the Institute for AI Policy and Strategy, noted that the report's lack of detail makes it impossible to verify if Google is living up to its public commitments, thereby making it difficult to assess the safety and security of its models.

Thomas Woodside, co-founder of the Secure AI Project, welcomed Google's decision to release a report for Gemini 2.5 Pro but expressed concerns over the company's commitment to delivering timely supplemental safety evaluations. He pointed out that the last time Google published the results of dangerous capability tests was in June 2024, for a model announced in February that same year. Woodside hopes that Google will start publishing more frequent updates, including results of evaluations for models that haven't been publicly deployed yet, as they could also pose serious risks.

Google's lack of transparency is not an isolated incident. Meta recently released a similarly skimpy safety evaluation of its new Llama 4 open models, and OpenAI opted not to publish any report for its GPT-4.1 series. This trend has raised concerns over the industry's commitment to AI safety and transparency.

Google's assurances to regulators to maintain a high standard of AI safety testing and reporting have added to the scrutiny. Two years ago, the company promised to publish safety reports for all "significant" public AI models "within scope." Similar commitments were made to other countries, pledging to "provide public transparency" around AI products. However, the recent report has fallen short of these promises, leading to concerns over the company's accountability.

Kevin Bankston, a senior adviser on AI governance at the Center for Democracy and Technology, described the trend of sporadic and vague reports as a "race to the bottom" on AI safety. He noted that combined with reports of competing labs like OpenAI shaving their safety testing time before release from months to days, this meager documentation tells a troubling story of a rush to market at the expense of AI safety and transparency.

Google has maintained that it conducts safety testing and "adversarial red teaming" for models ahead of release, although these details are not included in its technical reports. The company's spokesperson has promised that a report for Gemini 2.5 Flash, a smaller, more efficient model announced last week, is "coming soon." However, the lack of transparency and accountability in the AI industry has sparked concerns over the potential risks associated with these powerful models.

As the AI industry continues to evolve, it is essential for companies like Google to prioritize transparency and accountability in their safety reporting. The lack of detail in the Gemini 2.5 Pro report has raised questions over the industry's commitment to AI safety, and it remains to be seen whether Google and its rivals will take steps to address these concerns and provide more comprehensive safety evaluations in the future.