New assessment of attacks on Classic McEliece
Researchers analyzed recent progress in structural attacks on binary Goppa codes.
Recent structural attacks on McEliece with binary Goppa codes have progressed more in two years than in the previous four decades. These include a subexponential and a quasi-polynomial distinguisher. Key recoveries occurred for public challenges with brute-force labels up to .
Conditional key-recovery estimates range from about to per run of a large sparse linear-algebra computation. These estimates depend on the parameter set and how memory access is charged. On September 2026, the German BSI advised against new deployments of the scheme.
The analysis separates proved distinguishers, conditional cost estimates, and actual recoveries. It examines Saarinen's HOVER algorithm and his recovery of the TII-254 challenge. It also looks at Weis's cost study of Holdout/Hermite key recovery.
The cost model is the main quantitative uncertainty. There are about thirty bits between the cheapest and most expensive model for the same computation. For the four larger parameter sets, every model leaves conditional estimates far below the targeted levels.
For mceliece348864, the shortfall is the order of the spread between models. Structural estimates are lower than Grover-accelerated information-set decoding for the four larger sets in essentially every model. These are not demonstrated attacks because the underlying heuristics were tested on instances tens of thousands to hundreds of thousands of times smaller than the target.
Key facts
- Conditional key-recovery estimates are about to per run.
- The German BSI advised against new deployments of the scheme on September 2026.
- The cost model uncertainty is about thirty bits between the cheapest and most expensive model.