AI Could Speed Attacks on Cryptography, Debate Says
Projections of faster attack algorithms remain unproven, and specialists report no consensus that elliptic-curve cryptography will fall.
Updated on Nov. 3, 2026 in Artificial Intelligence
Ethereum co-founder Vitalik Buterin said AI could speed the discovery of more efficient cryptographic attacks, naming the elliptic-curve discrete logarithm problem, bilinear pairings and lattice cryptography as areas of concern. Coinbase's head of cryptography, Yehuda Lindell, responded that the cryptography community has no consensus that elliptic-curve cryptography will be broken. This is a claim and a projection, not a demonstrated result.
Why it matters
The debate concerns the pace of research that could yield faster attack algorithms against the math securing much of today's digital infrastructure, not a broken system. Buterin also argued that limits based on how AI is used are less harmful than limits based on users' identities.
The figure cited is a projection that AI could produce the equivalent of more than 50 years of human mathematical research within five years, and no measured benchmark or demonstrated attack is reported. The source does not establish how the estimate was produced, and no evidence supports the conclusion that elliptic-curve cryptography will be broken.
The players
Vitalik Buterin
Ethereum co-founder who said AI could speed the discovery of more efficient cryptographic attacks.
Yehuda Lindell
Coinbase's head of cryptography who said the cryptography community has no consensus that elliptic-curve cryptography will be broken.
The details
Elliptic-curve cryptography is a family of public-key methods that protects keys using the mathematics of points on a curve, and lattice cryptography relies on the difficulty of problems in high-dimensional grids. Bilinear pairings are mathematical maps used in some advanced schemes. Buterin's concern is that AI could accelerate the search for faster algorithms that solve these underlying problems, a line of research rather than an attack that exists today.
Timeline
Buterin's projection that AI could produce 50 years of research in five years was made in November 2026.
The Tech Race
The debate places AI-assisted mathematics against the long-standing security assumptions of elliptic-curve cryptography, bilinear pairings and lattice cryptography. The contrasting views come from a cryptocurrency co-founder and a cryptography head at Coinbase, with no consensus among specialists reported.
The projection does not change how existing encryption works for users today, and no timeline for any practical attack is announced. Readers should treat the claim as a research debate rather than an immediate security change.
The takeaway
Watch for published cryptanalysis results that demonstrate a concrete speedup against elliptic-curve or lattice problems, since such a paper would move the debate from projection to evidence. Until then, the specialist view that no consensus exists is the current state of the field.
Further reading
For broader context on how these tools are reshaping research, see the Artificial Intelligence section.
Source note: This article includes information reported by TokenPost.
