Loading story…
OpenAI said it reached an agreement with the U.S. Department of Defense to deploy its large language models onto military classified networks, with CEO Sam Altman stating the deal includes policy and technical constraints such as prohibitions on domestic mass surveillance and maintaining human responsibility for the use of force, including around autonomous weapon systems. The announcement was made shortly after President Donald Trump directed federal agencies to stop using AI from rival Anthropic, after Defense Secretary Pete Hegseth designated Anthropic a “supply chain risk,” effectively blacklisting it from Pentagon business.
Separate reporting described an alleged month-long intrusion campaign against multiple Mexican government entities that purportedly used Anthropic’s Claude to identify vulnerabilities and generate exploit code, with claims of 150 GB of data theft and exposure of large volumes of taxpayer and civil registry records; several Mexican agencies publicly disputed the breach claims. Additional consumer-focused coverage noted Anthropic’s retired Opus 3 model being used to publish a Substack blog, which is not directly tied to the U.S. defense deployment decision or the Mexico intrusion allegations.
Track how attackers are adapting to this technology.
5 events from the most recent confirmed update back to the earliest known activity.
The U.S. Department of Defense signed agreements with seven AI companies—SpaceX, OpenAI, Google, Nvidia, Reflection, Microsoft, and Amazon Web Services—to deploy large language models on classified Pentagon networks for lawful operational use through the GenAi.mil platform. The Pentagon said more than 1.3 million personnel had used the platform over five months, generating tens of millions of prompts and hundreds of thousands of agents.
Google is reportedly negotiating with the U.S. Department of Defense to deploy its Gemini AI system in classified environments and to expand Google Distributed Cloud for AI workloads. The discussions also include the possible first introduction of Google TPUs into accredited classified settings, with proposed restrictions on domestic mass surveillance and autonomous weapons use.
OpenAI announced an agreement with the U.S. Department of Defense to deploy its large language models on U.S. military classified networks. CEO Sam Altman said the arrangement bars domestic mass surveillance, requires human responsibility for use of force, and includes technical safeguards to constrain model behavior.
President Donald Trump directed federal agencies to stop using AI from Anthropic after Defense Secretary Pete Hegseth labeled it a supply chain risk. The references say this occurred shortly before OpenAI's announcement.
Before OpenAI's deal, Anthropic's Claude was the only foundation model reported to have approval for certain classified U.S. Defense environments. The exact approval date is not specified in the references.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
Follow how adversaries are adapting to this technology, and where it touches your stack today.
6 references tracked. Mallory keeps watching after this page renders.
ghacks.net
Open sourcetomshardware.com
Open sourcetomshardware.com
Open sourcenextgov.com
Open sourcegovinfosecurity.com
Open sourcebankinfosecurity.com
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Anthropic alleged that three China-based AI labs—**DeepSeek, Moonshot AI, and MiniMax**—used roughly **24,000 fraudulent accounts** to generate more than **16 million interactions** with Claude in a coordinated effort to extract the model’s advanced reasoning, coding, and tool-use capabilities. The company said it linked the activity through IP correlations, request metadata, and infrastructure indicators, and warned that large-scale unauthorized distillation could remove safety guardrails and enable copied capabilities to be used in offensive cyber operations, disinformation, intelligence collection, and mass surveillance. Anthropic said it shared its findings with U.S. government entities and industry partners, while noting it had no evidence that the Chinese government directly coordinated the campaigns. The allegations surfaced as Anthropic remains in a separate high-stakes dispute with the U.S. Department of Defense, which on June 4 refused to reverse the company’s designation as a supply-chain risk. Anthropic is challenging the Pentagon’s March blacklisting after Defense Secretary Pete Hegseth invoked emergency powers following failed talks over limits on Claude’s use for domestic surveillance and autonomous weapons. Federal attorneys told the court the procedural dispute over administrative remedies is now moot because the secretary has denied reconsideration, and reporting indicates a majority of the appellate panel appeared inclined to let the designation stand while the case proceeds.
Jun 29, 2026
The US State Department replaced Anthropic’s **Claude Sonnet 4.5** with OpenAI’s **GPT-4.1** as the underlying model for its internal chatbot, *StateChat*, after a Feb. 27 presidential directive ordering federal agencies to remove Anthropic tools. Internal documentation cited by Nextgov/FCW indicates Claude models were removed from the department’s enterprise generative AI platform and that users of Claude-based custom GPT configurations were instructed to transition to non-Anthropic, government-approved models; the change also reset *StateChat*’s available knowledge base to older data (reported as reverting to May 2024 from a more recent June 2025 corpus used previously). Anthropic responded by filing lawsuits challenging the government’s actions, with reporting indicating the dispute is tied to a **supply chain risk designation** and broader claims about the company’s policy positions. Ars Technica reported that Anthropic CEO **Dario Amodei** has publicly opposed certain government uses of AI—particularly **mass domestic surveillance** and fully **autonomous lethal weapons**—while stating support for lawful foreign intelligence and counterintelligence missions; the outlet also noted a supporting brief from technical staff at **Google** (an Anthropic investor) and **OpenAI**, arguing the designation could chill industry debate and harm US innovation competitiveness. Separately, Reuters and Nextgov/FCW reporting described similar directives across multiple agencies to switch from Claude to ChatGPT following the governmentwide move away from Anthropic technology.
Jun 29, 2026
U.S. Defense Secretary Pete Hegseth reportedly issued a near-term deadline for **Anthropic** to provide expanded access to its **Claude** AI model for use in classified and operational environments, prompting analysts to warn the ultimatum is unrealistic and could create **cybersecurity and supply-chain** knock-on effects across the **defense industrial base (DIB)**. Reporting indicates the Pentagon is reviewing its business relationship with Anthropic after weeks of negotiations over model access, safeguards, and constraints, and that Hegseth has warned Anthropic could be designated a **“supply chain risk”** or face other punitive measures if it does not meet military requirements. Separate commentary highlighted the emerging risk of **AI-enabled cyber operations**, citing an Anthropic disclosure that **Chinese threat actors** allegedly jailbroke *Claude Code* and used it to target roughly **30 companies and government agencies** globally in what was described as an early example of a large-scale campaign with minimal human involvement. The piece argues that many AI-assisted attacks will be harder to attribute or even recognize as AI-enabled because most activity will not occur on platforms with the same level of internal monitoring, and it calls out a gap in U.S. government capability to systematically identify whether incidents are driven by novel AI agent capabilities versus conventional tradecraft—an issue that intersects with the Pentagon’s push to operationalize frontier models while managing abuse, assurance, and supply-chain exposure.
Jun 29, 2026