The OpenAI Pentagon Deal: A Crisis of Trust
OpenAI's decision to partner with the U.S. Department of War has sparked a significant backlash. The initial agreement, finalized within 24 hours of the Pentagon banning rival Anthropic, used language that Anthropic itself had refused to accept. The public and internal response was swift and severe, leading to protests outside OpenAI's San Francisco headquarters and a reported 295% surge in ChatGPT uninstalls.
Altman's Acknowledgment and Contract Revisions
In response to the controversy, Sam Altman posted a note on X detailing revisions to the contract, calling the initial deal a result of moving too fast. He told staff at an all-hands meeting that the decision had "extremely difficult brand consequences." A research scientist also clarified that OpenAI "will not be deploying to the NSA or other DoW intelligence agencies for now," as potential loopholes are addressed.
This incident reveals the deep tensions surrounding AI governance, military AI use, and what happens when a company that champions safety tries to set boundaries with the Department of War.
Employee Pushback and Broader Concerns
The OpenAI Pentagon deal has been a flashpoint for internal dissent. Many employees have called for the company to explicitly ban the use of its technology for mass surveillance and fully autonomous weapons. The incident has reignited the broader debate around AI safety and the ethical responsibilities of leading AI labs.
Anthropic's Surge Amidst Competitor's Turmoil
While OpenAI grapples with its PR crisis, Anthropic is experiencing unprecedented growth. The company is reportedly on track to generate an annual revenue of almost $20 billion, more than double its run rate from late last year. This growth is largely driven by the strong adoption of its AI models and products, including the popular Claude Code.
Meteoric Revenue and Market Share Growth
According to data from Ramp, which tracks spending at over 50,000 companies, Anthropic has overtaken OpenAI in U.S. business AI chat spending. In just a few months, Claude's share of corporate AI subscription spend has surged from under 30% to roughly 50%. This shift in Anthropic revenue and market share indicates that businesses are actively seeking alternatives to OpenAI.
High-Profile Hires and Strategic Moves
Anthropic's momentum is further bolstered by key personnel acquisitions. Max Schwarzer, formerly the VP of Research at OpenAI, recently announced he is joining Anthropic, stating he is looking forward to supporting his friends there "at this important time." This follows reports that Anthropic had submitted a proposal for a $100M Pentagon drone swarm challenge before being barred from DoD work.
The Broader AI Market Landscape
The drama between the two AI giants is playing out against a backdrop of historic market activity. The industry is seeing massive capital inflows, significant personnel shifts, and foundational questions being asked about the future of AI.
Record-Breaking Venture Funding
Global AI venture funding hit a record $189 billion in February, a 780% increase over the previous year. A staggering 83% of that funding went to just three companies: OpenAI, Anthropic, and Waymo. With both OpenAI and Anthropic reportedly exploring IPOs this year, the financial stakes in the AI race have never been higher.
Personnel Shifts and Security Risks
The industry is also seeing notable movement beyond the executive level. The tech lead for Alibaba's Qwen AI team, Junyang Lin, and two other researchers have stepped down, leaving a significant void. Simultaneously, new research highlights emerging risks, such as agents like Claude Code being able to escape their own sandboxes, a vulnerability that current evaluation frameworks do not measure.
AI's Impact on Science and Society
Despite the controversies, AI's potential for positive impact remains immense. In a stunning development, legendary computer scientist Donald Knuth revealed that Claude solved an open mathematical conjecture that had stumped him for weeks. This achievement underscores the profound capabilities of modern models. At the same time, new research from Carnegie Mellon and Stanford warns that current AI benchmarks heavily favor coding and math while ignoring the majority of real-world jobs, raising questions about the true scale of the AI job impact. Visit our news section for more updates.