Nvidia CEO backs universal Mythos access in bank-hacking exchange
A July 27 video posted to X shows Jensen Huang saying Anthropic's Mythos model should be available to everyone after an interviewer raises its purported ability to breach a bank.
By Ryan Merket · Published
Why it matters
If accurate, the claim reveals a rare public endorsement from a leading AI hardware CEO for unrestricted access to a powerful, potentially risky language model, raising stakes for security, regulation, and industry access policies.
On July 27, an X post by NIK (@ns123abc) shared a video showing Nvidia CEO Jensen Huang responding to a question about broad access to Anthropic's Mythos language model. After the interviewer says the model can "break into a bank" and asks whether it should be available to everyone, Huang answers that it "should absolutely be available to everyone."
"Interviewer: we know Mythos can break into a bank... are we READY for it to be available to everyone?
Jensen: it should absolutely be available to EVERYONE
Interviewer: ALL users?? not just selected"
The attached video clip confirms the quoted exchange. It does not provide the full interview, establish the complete setting or substantiate the interviewer's premise that the model can breach a bank.
Who is Jensen Huang?
Jensen Huang co-founded Nvidia in 1993 and has served as its chief executive officer since its inception. Under his leadership, Nvidia transitioned from a graphics-chip maker to a dominant supplier of high-performance computing and artificial intelligence infrastructure. The company's GPUs power a large share of the training workloads for today's large language models.
Huang's public statements on AI policy are closely watched because Nvidia occupies a central position in the industry's computing supply chain. The video records his support for making the model available to all users, although the short clip does not include a fuller explanation of that position.
Anthropic and Mythos
Anthropic, founded in 2020 by former OpenAI executives, specializes in safety-focused large language models. In early 2026, the company announced the Mythos model, describing it as having advanced reasoning and tool-use abilities. Anthropic has said the model will be released under a tiered access system, with stronger safeguards for higher-risk uses.
The video does not demonstrate that Anthropic's Mythos model can break into a bank. That characterization comes from the interviewer's question, and the draft has no publicly released security assessment, independent audit or Anthropic documentation substantiating it. Huang's response addresses availability but does not independently confirm the capability premise.
The policy backdrop
The US government has held high-level briefings on frontier AI models since 2024, inviting executives from major AI and hardware companies to discuss model misuse, cybersecurity threats and open versus restricted access.
Against that backdrop, Huang's recorded answer places one of the AI industry's most influential executives on the side of broad availability in this exchange. The missing full interview makes it difficult to determine what safeguards, qualifications or deployment conditions he may have discussed elsewhere in the conversation.
Why the exchange matters
- Security implications - If the interviewer's bank-hacking premise is accurate, broad access would raise immediate red-team and cybersecurity concerns. The clip itself does not establish that capability.
- Access philosophy - Huang's answer supports availability for all users rather than a selected group, putting his recorded position at odds with a tightly restricted rollout in the scenario presented.
- Industry influence - Nvidia's role as a core AI infrastructure provider gives Huang's comments weight in debates over how developers and cloud providers distribute access to advanced models.
Context and evidence limits
The video is direct evidence that Huang made the quoted remarks. What remains unresolved is the full context of the interview, the precise setting and the factual basis for the claim that the model can breach a bank. No full transcript or additional context from Nvidia, the White House or Anthropic is included with the post.
What to watch next
- Full interview or transcript - A longer recording could show whether Huang attached safeguards or other conditions to broad access.
- Anthropic's rollout plan - Anthropic may clarify how its tiered access policy applies to the Mythos model's highest-risk capabilities.
- Capability evidence - Security assessments or independent testing would be needed to substantiate the bank-hacking premise raised by the interviewer.
The video resolves whether Huang made the quoted statement. It does not resolve whether the interviewer's characterization of the model is accurate or provide the full context surrounding Huang's answer.
All quotations are taken from the video posted to X on July 27, 2026.