AI Signal 418
OpenAI's Astra Solved Decades-Old Math Problems For $2,000
↗
The claim is sourced to a single feed (Slashdot, summarizing a Forbes report) and ultimately to OpenAI itself, so it should be read as a vendor announcement rather than independent verification. If the proofs hold up under community review, the interesting shift is structural: machine-checkable certificates attached to AI-generated mathematics at this scale, and a public claim that frontier reasoning models can displace long-standing open problems in active subfields rather than only reproduce known results.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
The work covers ten open problems across group theory, high-dimensional geometry, coding theory, quantum complexity, lattice cryptography, and extremal combinatorics, including a disproof of Connes's rigidity conjecture and an improvement to a 1978 sphere-packing bound.
OpenAI is reported to have published a 249-page manuscript collection on August 1, with a machine-checkable certificate attached to every claimed result, which is unusual for an AI math claim at this scale.
The story is carried by a single feed (Slashdot, citing Forbes) and originates from OpenAI's own announcement, so the proofs await independent review before the headline cost figure of $2,000 is treated as more than a marketing point.
THE CLUSTER