NEW YORK — US artificial intelligence lab Anthropic topped a new global AI safety ranking, but a report released on Tuesday warned that the industry as a whole is still falling short when it comes to tackling “existential” risks posed by advanced AI.
Meta climbed two places to fourth in the rankings, while xAI dropped three spots to seventh, just ahead of China’s DeepSeek and France’s Mistral, which finished last, AFP reported.
The rankings were compiled by the US-based AI safety think tank Future of Life Institute, which assessed nine of the world’s leading AI companies.
Seven researchers and governance experts evaluated the companies using publicly available information as well as data provided by the firms themselves.
They assessed performance across six categories: risk assessment, current harms, safety frameworks, existential safety, governance and accountability, and information sharing.
No company received an “A” grade in any category.
Anthropic achieved the highest overall score with a “C+”.
Mistral, which appeared in the rankings for the first time, disputed the assessment.
Asked by AFP about its last-place finish, the French company said the report’s framework was not suited to its approach to developing AI.
Unlike rivals such as Anthropic, OpenAI and Google DeepMind, which mainly develop closed AI systems, Mistral focuses on open models that users can download and modify.
“I was very disappointed to find that they came last, especially since Europe has really…been a leader in AI safety,” said Max Tegmark, an MIT professor and president of the Future of Life Institute.
“We reached out many, many times,” Tegmark added, saying Mistral did not respond to the organisation’s survey.
According to the institute, Alibaba, xAI and DeepSeek also declined to participate.
Three Chinese AI developers that produce open models also ranked in the lower half of the list: DeepSeek in fifth, Alibaba Cloud in sixth and Z.ai in eighth.
The report said several companies that had previously pledged not to allow their technology for military use had since softened those positions.
It specifically criticised Anthropic over what it described as “questionable military engagements”.
According to various media reports, the US government used Anthropic’s technology during military operations involving Venezuela and Iran over the past year.
The company was also recently barred by the Pentagon over disagreements related to AI safety, though that restriction was later lifted.
The report concluded that all nine companies remain inadequately prepared to address “existential” threats associated with artificial general intelligence (AGI), or AI systems capable of matching human-level intelligence.
While researchers acknowledged that “constructive attempts exist”, they said industry-wide efforts remain “entirely inadequate”.
The report also highlighted concerns that advanced AI could be misused to conduct cyberattacks or carry out tasks harmful to people.
Anthropic recently attracted attention after unveiling its most powerful AI model to date, Mythos.
The San Francisco-based company initially made the model available only to a small group of trusted organisations in April because of concerns its cyber capabilities could be exploited by malicious actors.
On June 12, the US government blocked Anthropic from releasing Mythos to foreign users on national security grounds.
The Trump administration lifted the restriction on June 30.
