Originally reported by Schneier on Security, WIRED Security
TL;DR
The Trump administration severed Pentagon ties with Anthropic over the company's refusal to support mass surveillance and autonomous weapons, while CBP used commercial advertising data to track phone locations. Iran has shut down 99% of internet connectivity amid ongoing air strikes.
Iran's 99% internet connectivity shutdown represents a massive state-level censorship event affecting millions, while CBP's use of commercial ad data for location tracking demonstrates systematic surveillance overreach by US agencies.
The Trump administration terminated the Pentagon's AI partnership with Anthropic after the company refused to modify its models for mass surveillance and fully autonomous weapons systems. Defense Secretary Pete Hegseth dismissed Anthropic's ethical guardrails as "woke," prompting a Friday evening order from Trump to discontinue all federal use of Anthropic models.
OpenAI immediately filled the void, securing potentially hundreds of millions in government contracts by agreeing to provide AI capabilities for classified systems. The move positions OpenAI CEO Sam Altman in the contradictory stance of claiming to uphold the same safety principles that led to Anthropic's dismissal.
The administration escalated beyond contract termination, designating Anthropic as "a supply-chain risk to national security" - a classification previously reserved for foreign entities. This designation blocks not only direct government contracts but also prevents federal contractors and suppliers from working with the company. Officials have threatened to invoke the Defense Production Act to force modifications to Anthropic's safety protocols.
As Schneier and Sanders note in their analysis, this dispute highlights the commoditized nature of current AI models, where top-tier offerings perform similarly and branding becomes a key differentiator. Anthropic's positioning as the "moral and trustworthy" AI provider may prove more valuable than the lost Pentagon contracts, while OpenAI risks politicizing its brand through association with controversial military applications.
U.S. Customs and Border Protection used location data harvested from online advertising networks to track mobile phone movements, according to new reporting from WIRED. The practice represents another example of federal agencies circumventing traditional surveillance oversight by purchasing commercially available data that would otherwise require warrants.
The technique leverages the sprawling data broker ecosystem that collects precise location information from mobile apps and websites, then packages it for sale to advertisers. By purchasing this data rather than obtaining it through legal channels, CBP avoided the judicial oversight typically required for location tracking.
This revelation adds to mounting evidence of federal agencies exploiting the commercial data marketplace to conduct surveillance operations outside established legal frameworks. The practice raises significant Fourth Amendment concerns about unreasonable search and seizure in the digital age.
Iran has reduced national internet connectivity by 99 percent in what appears to be one of the most comprehensive digital blackouts ever implemented by a state actor. The shutdown coincides with ongoing air strikes that have likely caused additional infrastructure damage beyond deliberate government restrictions.
The near-total connectivity loss leaves Iranian citizens with severely limited communication options and virtually no access to global information networks. Traditional circumvention tools that helped maintain some connectivity during previous Iranian internet restrictions have proven largely ineffective against this comprehensive approach.
The scale of the shutdown demonstrates the Iranian government's willingness to sacrifice economic activity and international connectivity to maintain information control during a period of military action. The move effectively isolates the population from external communications and limits their ability to document or report on domestic events.
Bruce Schneier's weekly "Friday Squid Blogging" featured an unusual historical parallel to modern security policy gaps. Byzantine monks at Constantinople's Monastery of Stoudios operated under strict dietary regulations that explicitly prohibited meat, dairy, eggs, and restricted fish to feast days. However, squid remained permissible due to taxonomic confusion rather than intentional policy.
The squid's eight arms, boneless structure, and color-changing abilities defied easy categorization in medieval regulatory frameworks. Rather than creating specific rules for ambiguous creatures, monastery administrators simply allowed squid consumption to continue unchallenged.
This historical example mirrors modern cybersecurity challenges where new technologies often fall into regulatory gray areas until specific policies develop to address them. The tendency for unusual or undefined elements to persist through administrative oversight rather than explicit approval remains relevant to contemporary security policy development.
Originally reported by Schneier on Security, WIRED Security