Hope for a benevolent AI: Musk joined Amodei's discussion on risks and regulation
The technology community's attention has once again turned to the issue of artificial intelligence safety. Elon Musk unexpectedly supported Anthropic CEO Dario Amodei in his recent public statement about the risks associated with AI development. Musk posted a response almost immediately after Amodei left a comment saying "Interesting dialogue" under a post, defending his predictions against accusations of excessive pessimism.
Why does Amodei insist on AI risks?
The discussion was sparked by investor Gavin Baker, who noted that rumors about Amodei spread quickly because his public statements rarely diverge from his actual position. Amodei himself rejected accusations that he only talks about dangers. According to him, the company's two main essays divide attention between risks and benefits equally. Short interviews, as he explained, often look gloomy because such formats gather more clicks but do not reflect the full picture.
Amodei also named the deep reason for distrust: ordinary people do not trust corporations, government agencies, or the IT industry. This distrust automatically extends to artificial intelligence, accumulating over decades. That is why in June he proposed mandatory testing of leading models instead of voluntary promises, and a few days later appealed to the G7 with a warning about the danger of fragmented national regulations.
"The fairest criticism of AI companies, including Anthropic — we still haven't fulfilled the main promise: to bring benefit to the world," Amodei stated.
Musk and the philosophy of "benevolent" AI
Musk called the dialogue interesting and emphasized the complexity of the debate. Beyond regulation, the discussion touched on philosophical questions. Naval Ravikant framed them as follows: you cannot create a god and put it on a leash. Musk almost verbatim repeated this thought: "I hope AI will be kind to us." His restrained admission carries a warning: if people lose control over superintelligence, society will have no choice but to rely on the goodwill of its own creation.
Earlier, Musk had already criticized OpenAI's closed nature and noted that Amodei had chosen a different path. Notably, the Anthropic head also challenged the view common in Silicon Valley, pointing out that regulation does not always lead to a concentration of power. California's SB 53 AI safety law, which Anthropic supports, exempts companies with revenue below $500 million from requirements.
Amodei named two approaches he considers correct. The first was proposed by the Trump administration — testing advanced models before release. The second was put forward by Google DeepMind head Demis Hassabis — creating a regulator modeled on FINRA. In conclusion, Amodei returned to his most prominent promise: AI in biology could cure most human diseases within five to ten years. At Anthropic and Pfizer, AI is already considered key infrastructure in medicine, and the CEO himself linked this goal to a personal story — his father died of hepatitis C shortly before effective treatments became available.
My comment: It is telling that the discussion about AI safety is increasingly moving from the technological sphere into the sphere of trust and philosophy. For the market, this is an important signal: regulatory uncertainty remains the main risk factor for AI-related projects, and investors should consider not only technical metrics but also the political context surrounding the largest players.