National Security

The U.S. government, through the Center for AI Standards and Innovation (CAISI), is intensifying scrutiny of advanced AI systems by forming partnerships with tech giants like Google DeepMind, Microsoft, and xAI. These agreements enable pre-deployment evaluations of frontier AI models to assess capabilities and enhance security. This proactive oversight initiative is part of a broader governmental push, with discussions underway for a potential AI working group to explore regulatory frameworks. The aim is to balance innovation with safety, mitigating risks from powerful AI models before public release.

The DOD faces a complex challenge with Anthropic’s AI model, Mythos, due to its advanced cyber capabilities, even as Anthropic itself is a designated supply chain risk. While the Pentagon has flagged Anthropic’s Claude models, preventing their use by defense contractors, Mythos presents a separate national security concern. This necessitates robust network fortification. Despite ongoing legal disputes and the supply chain risk designation, the DOD is reportedly exploring ways to leverage Mythos, while also formalizing agreements with other AI leaders.

True Anomaly secured $650 million in Series C funding, valuing the space defense startup at $2.2 billion. The capital will scale operations, expand its workforce to 500, and accelerate development of advanced space interceptors. This investment highlights growing interest in commercial space defense, driven by geopolitical tensions and government initiatives. True Anomaly aims to significantly increase its manufacturing footprint and launch new products, bolstering its role in the burgeoning space interceptor market.

China’s NDRC has ordered Meta to divest its $2 billion acquisition of Singapore-based AI startup Manus, citing national security and regulatory compliance. This move targets the “Singapore-washing” trend of Chinese tech companies relocating offshore and underscores Beijing’s efforts to control its burgeoning AI sector amidst rising geopolitical tensions. The decision signals increased scrutiny on cross-border AI investments.

A D.C. appeals court denied Anthropic’s emergency request to halt the Pentagon’s blacklisting of its AI models. The court found Anthropic’s concerns to be primarily financial, outweighing them against the government’s national security interests in securing vital AI technology during conflict. This ruling contrasts with a separate San Francisco injunction blocking a Trump-era ban on Claude AI. The battle highlights the complex regulatory landscape for AI in defense.

Prediction markets face intense scrutiny following Polymarket’s removal of a forum betting on a military rescue. Critics condemn the ethical implications of such speculation on human lives. Legislators are pushing for stricter regulations, prohibiting bets on elections, wartime events, and deaths, citing national security risks. The CFTC is asserting its regulatory authority, while organizations like the NFL request operators avoid “objectionable bets.” These developments highlight the evolving challenges of balancing innovation with public trust and responsible oversight.

The UK is leveraging advanced AI, like Palantir’s platform, to enhance financial oversight and national security. The FCA’s pilot program uses AI to detect money laundering, insider trading, and fraud. In defense, AI aids military decision-making and targeting. Stringent data protection controls are in place, with vendors acting as data processors and data hosted domestically, ensuring privacy and control over sensitive information.

The Pentagon has labeled Anthropic’s AI models as a “supply chain risk” due to concerns over embedded policy preferences compromising defense integrity. This unprecedented move requires contractors to certify non-use of Anthropic’s Claude models. Anthropic is suing the U.S. government, citing irreparable harm and jeopardized contracts. The Department of Defense states the action isn’t punitive, acknowledging a transition period and existing usage by contractors like Palantir.

OpenAI CEO Sam Altman’s visit highlighted tensions over AI in national security. Discussions with lawmakers focused on defense applications and OpenAI’s DOD agreement, emphasizing the need for safeguards and constitutional compliance. This follows Anthropic being designated a national security risk by the DOD over disagreements on AI use, particularly concerning autonomous weapons and surveillance. Altman stressed OpenAI’s safety principles, which the DOD reportedly accepted. Legislation is being drafted to establish guidelines for DOD AI contracts, underscoring the crucial role of congressional oversight in navigating rapid technological advancements.

OpenAI CEO Sam Altman clarified that the company does not control the Pentagon’s operational decisions regarding its AI technology. This follows a new DOD contract, sparking debate amid geopolitical tensions. Altman emphasized that while OpenAI provides technical input and safety protocols, the DOD retains ultimate control. This distinction highlights the complex ethical landscape of AI deployment in national security, with competitors like xAI adopting a more unconstrained approach, creating a bifurcated market in AI defense.