Overview
The Future of Life Institute (FLI), a U.S.-based nonprofit focused on reducing large-scale risks from transformative technologies, published its Summer 2026 AI Safety Index on July 7, 2026. <cite index="1-11,1-12">The index evaluates nine leading AI companies on 37 indicators spanning six critical domains; the nine companies include Anthropic, OpenAI, Google DeepMind, xAI, Z.ai, Meta, DeepSeek, Alibaba Cloud, and Mistral.</cite>
Methodology
<cite index="1-13,1-14">The index collected evidence up to June 3, 2026, combining publicly available materials—including model cards, research papers, and benchmark results—with responses from a targeted company survey. An independent panel of seven leading AI researchers and governance experts reviewed company-specific evidence and assigned domain-level grades (A–F) based on absolute performance standards with discretionary weights.</cite> <cite index="6-13">The six evaluated categories are: risk assessment, current harms, safety frameworks, existential safety, governance and accountability, and transparency and communication.</cite>
Scores and Rankings
<cite index="5-10">Evaluated by a panel of technical experts using the U.S. GPA grading system, the report shows that Anthropic, OpenAI, and Google DeepMind lead the industry, though no model developer scored higher than a C+.</cite> In detail: <cite index="6-9,6-10,6-11">Anthropic took the top overall spot with a C+; OpenAI and Google DeepMind each received a C, while Meta managed only a D+. Z.ai and Alibaba Cloud scored a D−, and xAI, DeepSeek, and France's Mistral all received the lowest possible grade of F. Not a single company earned an A or B.</cite>
<cite index="1-1">Anthropic again earned the highest overall grade and led five of six domains via relatively strong transparency, a comparatively established safety framework, technical research, and governance.</cite> <cite index="1-3">OpenAI now leads in the Risk Assessment domain on the strength of a broader evaluation suite and diverse engagement with external testing.</cite>
Among notable movements in the rankings, <cite index="1-4">Meta improved from 6th to 4th place, while xAI dropped from 4th to 7th place.</cite> <cite index="16-25,16-26">After placing fourth in the second half of last year, xAI plummeted to seventh place, with the decline attributed to a significant weakening of safety-related transparency and governance following its acquisition by SpaceX.</cite>
Key Findings
Existential safety is the weakest domain industry-wide. <cite index="1-18,1-19">No company exceeds C− in existential safety; most score D or below. Constructive attempts exist—such as Anthropic's constitutional classifiers, OpenAI's call for governance institutions, and Google DeepMind's monitoring commitments—but are judged by panelists to be "entirely inadequate."</cite>
Redline commitments are eroding. <cite index="1-16,1-17">Anthropic, OpenAI, Google DeepMind, and Meta have weakened or voided pledges to pause unilaterally if redlines are approached, some citing competitor-contingent conditions. Reviewers call this "moving goalpost" and argue that it has "undermined safety frameworks across the board."</cite>
Military AI use flagged as an emerging harm. <cite index="1-8">From 2024 to 2026, companies including Anthropic, OpenAI, Google DeepMind, and Meta that previously banned military applications gradually reversed course, joining xAI and Mistral in actively seeking defense contracts.</cite>
Safety progress has plateaued. <cite index="5-21,5-22">Both OpenAI and Google DeepMind followed a familiar arc, peaking in late 2025 before sliding back in the current report, as translating written safety frameworks into concrete operational practice is proving difficult for market leaders.</cite>
Geographic reach of failure. <cite index="1-7">Three companies received failing grades—one each from the U.S. (xAI), China (DeepSeek), and Europe (Mistral).</cite> <cite index="1-6">Although the European Union is a leader in AI safety regulation, the top European AI company, Mistral, scored dead last on safety.</cite>
Expert Reaction
<cite index="1-10">"While there is good work being done on AI safety in the industry, the capabilities race has become more extreme,"</cite> said Professor Stuart Russell of UC Berkeley, one of the panelists. FLI President Max Tegmark stated that <cite index="16-21,16-22,16-23">"AI companies are racing full speed toward a cliff. They warn about the dangers of artificial superintelligence themselves, yet they refuse to stop competing to build it. Voluntary commitments were never sufficient to begin with."</cite>
<cite index="1-9">The index collected evidence up until June 3, 2026, and does not include possible new recent events.</cite> FLI publishes the index twice yearly as part of its ongoing effort to benchmark industry safety practices against consistent metrics.