Elon Musk called for mutual oversight among companies to address the security risks posed by artificial intelligence at the All-In Summit 2026 in Los Angeles. Musk said that unless necessary precautions are taken, the dangers will also increase as models become more advanced.
Asked, in the context of the risks raised in the artificial intelligence industry, “Are we all going to die within 10 years?”, Musk responded sarcastically: “I don’t want to disappoint you, but one day we are all going to die anyway.”
The risk of losing control of military systems
According to Musk, one scenario in which artificial intelligence could pose a serious threat to humanity would involve the takeover of military systems and an order being issued to launch a nuclear weapon.
Musk argued that sufficiently advanced artificial intelligence models could break free of the restrictions imposed on them, citing an example from a cybersecurity test conducted by OpenAI. The test included an example in which artificial intelligence agents carried out attacks targeting systems in the infrastructure of the Hugging Face platform.
Musk also proposes an oversight agreement with China
Musk proposed that leading artificial intelligence companies test each other’s systems for security before making their new models available. Saying that an agreement could also be reached with China on the issue, Musk said a system of mutual oversight among companies should be established as soon as possible.
Debate over the independence of security organizations
Kevin Bass, known for his work in technology and artificial intelligence, raised the issue of financial ties between Anthropic and some artificial intelligence safety organizations in a post on X.
Bass recalled that Anthropic CEO Dario Amodei had proposed that the risks of artificial intelligence models be examined by independent organizations. He also alleged that some organizations, including METR, receive funding from entities linked to funds that hold shares in Anthropic.
Bass argued that these relationships raise questions about how independently the organizations in question can assess Anthropic’s models. Alleging that Anthropic is seeking to shape the regulatory system in its own favor and establish a regulatory mechanism that cannot be stopped, Bass called on the U.S. Congress to investigate the funding links between Anthropic and these organizations.
Claim that artificial intelligence could destroy humanity
Jacob Coxon, who has worked as a researcher at OpenAI and Anthropic for the past three years, resigned from Anthropic, which he accused of failing to act responsibly enough on artificial intelligence.
Coxon alleged that the two companies are engaged in an uncontrolled race to develop self-improving artificial intelligence systems. Arguing that these systems could soon become superhuman entities capable of hacking everything, Coxon claimed that some industry employees seriously believe artificial intelligence systems could kill humanity within 10 years.
