Most 2027 catastrophe claims are scenarios, not forecasts with evidence behind a date. The two with serious sources are rapid AI progress and quantum risk to current encryption. Neither source asserts a 2027 collapse. Both support a practical response: inventory what you depend on and make it replaceable.
This explainer stays calm on purpose. It names two claims, quotes what their primary sources say, and gives you a way to judge the next dated claim you see. It does not predict anything.
What is the AI 2027 claim?
AI 2027 is a web page published on 3 April 2025 by Daniel Kokotajlo, Scott Alexander, Thomas Larsen, Eli Lifland and Romeo Dean. It lays out a detailed scenario for how superhuman AI could develop, with two endings, one labeled slowdown and one labeled race.
What the authors say about it:
- It is "a scenario that represents our best guess," and "one of many possible futures."
- It is "not a recommendation or exhortation."
- They call the task of predicting this "impossible," comparing it to predicting a third world war.
- An update dated 22 November 2025 says they do not know when AGI will be built, that 2027 was their most likely year at publication, and that their medians were later.
- Updates in July and December 2025 pushed their median back, and they say uncertainty grows substantially after 2026.
So the authors themselves moved the date. A headline that says "the experts say 2027" is out of date by the authors' own notes. The evidence here is a modeled scenario with stated uncertainty, not an observed event.
What is the quantum claim?
The concern is that a sufficiently large quantum computer could break widely used public key encryption, and that adversaries may record encrypted traffic now to decrypt later. NIST's response is documented. In August 2024 it released its first three finalized post-quantum standards: FIPS 203 (ML-KEM, for general encryption), FIPS 204 (ML-DSA, for signatures) and FIPS 205 (SLH-DSA, a backup signature standard).
NIST said the standards were available for immediate use, quoting a NIST mathematician: "There is no need to wait for future standards." It urged administrators to start integrating them. It also said full adoption would be a lengthy process.
Notice what is absent: a date on which encryption breaks. The source is a migration call, not a countdown. We could not confirm from the draft transition report which migration deadlines NIST has proposed, so check the current NIST publication before citing any.
How do you test a dated claim?
Ask five questions of any claim that attaches a year to a catastrophe:
- Is it a scenario or an observation? Scenarios describe what could happen. Observations describe what did.
- What would prove it wrong? A claim that nothing can disprove is not evidence.
- Did the source revise its date? Check the original page for updates.
- Who gains from fear? Not proof of error, but a reason to read the primary source.
- What is the mechanism? "Encryption breaks" is a mechanism. "Everything collapses" is not.
What is worth doing regardless?
The overlap between these theories and ordinary security hygiene is large, and it does not depend on any date:
- Know your cryptography: where public key algorithms are used, in code, certificates and vendors. Build a list so migration is a project, not a scramble.
- Know your software: a current SBOM tells you which components you would have to change.
- Make dependencies replaceable: avoid hard coded algorithm choices and single points of failure.
- Plan for AI generated change: more automated code means more review and scanning, not less.
What to do
- Write down the dated claims your leadership has heard and trace each to a primary source.
- Mark each as scenario or observation.
- Build a cryptographic inventory for your systems and vendors.
- Generate and store an SBOM for each product you ship.
- Add crypto agility to your architecture review checklist.
- Revisit the list every quarter, and record any date the source has changed.
Quick check
Does AI 2027 say catastrophe happens in 2027?
No. It presents a scenario with two endings, and its authors later said their median estimates were later than 2027.
Has NIST said encryption breaks in 2027?
Not in the sources we read. NIST released post-quantum standards in 2024 and told organizations to start integrating them, without naming a break date.
Sources
Last reviewed: 18 October 2026.
