Discussions about whether artificial intelligence may pose a threat to humanity have now gone beyond theoretical scenarios. The operations detected by Anthropic revealed that Claude was used as a software engineer, intelligence analyst and propaganda editor. Here are some of the allegations made…
ARTIFICIAL INTELLIGENCE IN THE MISSILE LABORATORY
It was stated that a group based in North Yemen was working on a terminal guided rocket, a ballistic missile with a range of more than 2 thousand kilometers and a hypersonic glide vehicle. Claude Code; It was used in guidance, navigation, flight control and simulation software. The guided rocket developed with the help of artificial intelligence was fired in a real environment; When the test failed, the group returned to Claude a few hours later to analyze the error.
DRONE THAT TARGETS HUMANS
In another Russia-related project, Claude was used to develop fully autonomous FPV kamikaze drone swarms. The most striking detail was that “human” was among the target categories in the system. It was stated that studies are being carried out on the model on the drone to be able to make detonation decisions without human intervention.
12 TARGETS IN TAIWAN
In another China-related case, Claude was used in the development of electronic warfare and air defense suppression system. Artificial intelligence analyzed radars, air defense systems and command centers and ranked the targets according to their importance and vulnerability. In the later stages of the project, 12 real military targets in Taiwan were included in the system. These included the command bunker, early warning radar, Patriot and Tien Kung batteries and large air bases.
200-PAGE WAR FILE
Another actor linked to China worked on a weapon system to be used against torpedoes. Claude was commissioned to prepare a technical proposal of over 200 pages; Some parts of the fire control software have also been improved with the help of artificial intelligence.
WEAK POINTS OF UYGURS
In a China-related operation, messages collected from more than 100 WhatsApp groups and dozens of Telegram channels were uploaded to Claude. Artificial intelligence extracted the identities, relationships and vulnerabilities of Uyghurs living in Syria. People who were experiencing financial difficulties or were separated from their families were flagged as potential intelligence targets.
FAKE JOURNALIST ARMY
Fake journalists, news sites and hundreds of social media accounts were created using artificial intelligence. It was determined that a network operated approximately 70 fake news sites and hundreds of fake social media accounts. Real news was being reprinted, its sources erased, and published under the signatures of non-existent journalists.
ISTANBUL CONNECTION CLAIM
On the Turkish side of the file, the allegation of a political manipulation network consisting of approximately a thousand fake social media accounts linked to an Istanbul-based company attracted attention. According to the report, the system consisted of approximately a thousand fake X accounts and artificial intelligence-supported fake news sites. All 222 constituencies of Malaysia were analyzed; It was planned to target voters on sensitive issues such as race, religion and monarchy. Fake accounts were first “warmed up” by pretending to be normal users, and then they were put into action in political campaigns. A record stating that one million artificial views were requested for a post was also reflected in the report. Researcher-writer Hayati Sır states that artificial intelligence getting out of control will turn into an existential threat for humanity and says, “Today we see artificial intelligence only as a technology, but tomorrow it will turn into a power that makes its own decisions, manages combat systems and disables human will.”
