SNL mocks Dario Amodei and exposes the paradox of building the AI you say could destroy us

By Zendoric · September 29, 2026.

Saturday Night Live devoted this weekend's Weekend Update to Dario Amodei, Anthropic's CEO, and the video circulated among people in the tech industry itself, not just among the show's audience. Actress Jane Wickline plays a nervous, sweaty Amodei with mannerisms evoking Gollum, in a parody built on two real events, as CNET reports: an interview Amodei gave Anderson Cooper on CNN in which the executive refused to rule out that AI could wipe out humanity, and his essay "We Must Pace the Frontier", published amid a surge of talk about human extinction, in which he called on the industry's companies —his own included— to slow development down.

The sketch's central joke is also its sharpest criticism. The fictional Amodei promises: "I want to assure you that, if we can pressure lawmakers into creating safeguards, we will be able to stop me." When host Michael Che asks him how his company will avoid killing us all, the character stammers: "That's a very hard question", before launching into a dialogue with himself in the voices of Sméagol and Gollum that ends with "AI is the devil, and I am its creator". The character adds that he and other industry executives —he explicitly names OpenAI's Sam Altman— "do not approve of what they are doing", and closes with the line that gives this piece its title: "AI is not a weapon. It's a tool. A tool for building weapons. And I'm asking you to ask me to stop." The video sparked debate on social media; among the comments CNET quotes, one joked that "we'll be revisiting this in our final days, in the next ten years". Anthropic did not respond to the outlet's request for comment.

That an AI capable of existential risk is the subject of a comedy sketch on mainstream television is no small detail. Barely a couple of years ago this kind of discourse was confined to X threads, safety papers or conference panels; seeing it land on Weekend Update is a sign that anxiety about controlling AI is now popular culture, not niche jargon. Comedy works here as a social thermometer: people laugh at something because it already weighs enough to require a release valve.

But the sketch gets something deeper right than the caricature: it puts its finger on the structural contradiction Amodei has long defended in public, one satire does not invent, only exaggerates. The "pacing" argument —slowing the pace— as it reads from the outside, is not that Anthropic will stop on its own: it is that, in a competitive race between several labs, no player can decelerate unilaterally without losing ground to the rest, so the only way out is for governments to impose the brakes on everyone at once. It is, literally, asking to be forced to do what you say you want to do. SNL sums it up better than any policy paper: "I'm asking you to ask me to stop" is an admission that responsibility is being outsourced to regulators who, as things stand, lag far behind the technical capability they claim to want to contain.

It is worth separating, however, two things the sketch deliberately mixes for laughs: an executive's public anguish about catastrophic risks is a posture, not evidence. Nobody has proven that AI will destroy humanity, just as nobody has proven the opposite; these are claims about the future, not verified facts, and they should be treated as such however much institutional weight sits behind them. That said, honesty about the short-term risks —loss of control over increasingly capable systems, concentration of power in a handful of labs, the absence of real governance— is exactly the part of the conversation that must not be sugarcoated, and there the SNL joke, exaggerated as it is, points to a legitimate problem: today, safety depends largely on the goodwill of the very companies competing to move faster.

Our reading is that this cultural vignette does not change the underlying bet: well governed, AI's trajectory still points toward a radical reduction in disease and an abundance of resources that today looks like science fiction. But that long-term bet only holds up if the "ask me to stop" phase translates into real safeguards —traceability, external audits, coordination between labs and governments— and does not remain a rhetorical gesture the industry itself repeats in interviews and essays while capex and model capability keep rising with no visible brake. That the joke landed hard enough to reach mainstream television says more about the collective mood toward AI than about the technology's actual destiny; the difference between the two is precisely what needs watching in the months ahead.