Anthropic: No "kill switch" for AI in classified settings
Quick Insights
The Bottom Line
Anthropic states it cannot control its AI models once deployed by the Pentagon, raising misuse concerns.
AI Summary
Anthropic stated in a new court filing that it has no "kill switch" or technical ability to control its AI models once deployed by the Pentagon. This filing was submitted to a federal appeals court in D.C., where the Pentagon designated Anthropic a supply chain risk. The Pentagon contends Anthropic is inappropriately involved in how its technology is used in sensitive military operations, despite the company's usage policies banning Claude for autonomous weapons. A D.C. court previously rejected Anthropic's request to pause the designation, while a California judge granted a similar request in a parallel case. A hearing for this dispute is scheduled for May 19.
What's Being Done
A federal appeals court in D.C. will hold a hearing on May 19 regarding the Pentagon's designation of Anthropic as a supply chain risk.
Source Coverage Map
3 of 43 tracked sources covered this story
Following this story?
Get notified when new coverage appears
Other Sources Covering This Story
3 sourcesMultiple outlets have reported on this story. Compare perspectives from different sources.
This article is part of a story we're tracking:
Should this be getting more attention?
You Might Have Missed
Related stories from different sources and perspectives
PoliticsCalifornia officials tout historic drop in violence as killings fall to record low
<p>State’s strict gun policies heralded as data shows 35% reduction in homicides between 2022 and 2024</p><p>California officials are touting a historic three-year decrease in homicides and gun violence that has led to the state’s lowest number of killings on record.</p><p>The number of homicides in California decreased by 35% between 2022 and 2024, with 2,304 deaths reported in 2022 and 1,768 in 2024, <a href="https://wonder.cdc.gov/controller/datarequest/D158;jsessionid=90709D2EC66131AC6C4E21AF4D29?stage=results&action=sort&direction=MEASURE_DESCEND&measure=D158.M1">according data from</a> the US Centers for Disease Control and Prevention. The dip in homicides was most pronounced for teenage and young adult Black and Latino males, who have historically faced the highest risk of being killed or injured by gunshot wounds. Suicides, the most common type of gun deaths, also fell to record lows, according to the report.</p> <a href="https://www.theguardian.com/us-news/2026/...
AI & WarfareUS security agency is using Anthropic's Mythos despite blacklist, Axios reports - Reuters
<a href="https://news.google.com/rss/articles/CBMivwFBVV95cUxPZVQxQUdjUjgtWnUzUnNfXzI5WFdiMXNkOC1qakZhZ0hfMEZvUVBpckJXQWlFTkdKTzljaU1vQm5hSDFOY3ZHc2p3M1pSSVp5enZuS0VfS3RiT3o4MWx4MnE2dkNHaWNDNVZlR0xUZDRpbEc5eW83V3BmQm9CX1JTeWxvdXJubnJwT19lTXJaLUotTzJ6X1ZER0poQ2N5VEMzOUNuWUxJZWxqYnlRVFNjQzJHeUhCalNNY21KMDJLSQ?oc=5" target="_blank">US security agency is using Anthropic's Mythos despite blacklist, Axios reports</a> <font color="#6f6f6f">Reuters</font>
AI & WarfareSteve Bannon sides with Anthropic in fight with Pentagon: ‘It’s almost too dangerous’
Former White House strategist Steve Bannon said he believes artificial intelligence company Anthropic “had it right” in demanding its technology not be used in fully autonomous lethal weapons, setting off a fight with the Pentagon. The Pentagon has insisted it be allowed to use Anthropic’s Claude for “all lawful uses” and last month labeled the…
AI & WarfareWhite House and Anthropic CEO discuss working together amid rising fear about Mythos model - Reuters
<a href="https://news.google.com/rss/articles/CBMimgFBVV95cUxOTGRzSHhnbEhCQjM5UWpyQllSNzBfRWhMdGFPQ2N2VmNDNVh1NFZoVnNqMEFyS19MZkZGTVBKelFuckZEc2x2Z2ZuVnFDallPR3Rmeng0em5tT1kzZC13YXhyZ0xZdlV2N2tJS1RhYVdEbjRPSWRpOTB2NVY1VWhqalZFRHpYMkNJZEs3Y0hXYmxvR2pENURpZnV3?oc=5" target="_blank">Anthropic CEO Dario Amodei arrives at White House for talks</a> <font color="#6f6f6f">Reuters</font>
National SecurityUS soldier charged with using classified intel to win $400K Polymarket bet on Maduro raid - AP News
<a href="https://news.google.com/rss/articles/CBMingFBVV95cUxQcjV2ajk3b0pGb3FfcEwxWEJMelJEcGJ1LWVnWjZqWXcxdExIdnU3cG5lOFFxRkVFSGNtcUpUOEw1YVgzT2pmeVRXaGZGYTlNd3llYVN6M294ejJOZDdGNGJqRVRQMkIyWWlKT1Nkck9iVDYtcVhBbFpLdEFoOVRVYmtKR1VIa2ExbU9HRk03YXNpd0J1RGVzdHdxNUhuUQ?oc=5" target="_blank">US soldier charged with using classified intel to win $400K Polymarket bet on Maduro raid</a> <font color="#6f6f6f">AP News</font>
Civil RightsEx-officer planned to kill Black people in mass shooting at a New Orleans festival, authorities say - AP News
<a href="https://news.google.com/rss/articles/CBMiowFBVV95cUxOY0lZNEVUYzlPOGVHMGl3cUtyZ2NTUEtkSlJQUzFOaHV6V3lmNWlON3BIaHBZSFFkeUhOcFpkVHpiS2MxdzYwZjhiZVdBOXc3TTlodFQzNUNvcFRQbHRQUzVVOUYtR0dKMVlDZEVTVXlGbWlSUUJGalR5SWREREtMaTdiV1NRTTl6Ul9NbjVjSnB5cGFqa3dqVFdkODg0ZDJoUmpn?oc=5" target="_blank">Ex-officer planned to kill Black people in mass shooting at a New Orleans festival, authorities say</a> <font color="#6f6f6f">AP News</font>
Florida investigates OpenAI over ChatGPT's alleged role in college shooting
OpenAI and ChatGPT are under investigation by Florida officials after a deadly shooting last year at Florida State University. Prosecutors allege the AI bot offered "significant advice" to the suspect just days before the shooting. OpenAI says its chatbot is not responsible. Jo Ling Kent reports.
Did this story change how you see things?
Stories like this only matter when people see them. Help us get verified journalism in front of more eyes.
The Verity Ledger curates verified investigative journalism from trusted sources only.
See our sourcesMost Read This Week
Report: Merck’s blockbuster cancer drug topped $200,000 a year under Trump
One in three Americans forced to make financial sacrifices for health coverage
Trump-branded data center project CEO departs
6 crew still missing after overturned ship that disappeared after typhoon is found near Saipan - AP News
US Live Nation and Ticketmaster verdict triggers calls for Australian investigation into ticketing rules
