Worldcoin (WLD) AI Watch: OpenAI Astra Solves 10 Math Problems

Worldcoin News

Worldcoin (WLD) is being watched by crypto participants as an altcoin exposed to the broader AI verification narrative, after OpenAI researcher Noam Brown disclosed on Aug. 1 that an internal version of the company’s next-generation model family, Astra, produced solutions to 10 mathematics and theoretical computer science problems that had seen no meaningful progress for at least a decade. The disclosure positioned Astra as a long-horizon reasoning system capable of coordinating multiple agents over hours or days, rather than answering prompts in a single pass. The problems spanned high-dimensional geometry, coding theory, group theory, quantum complexity, lattice cryptography and extremal combinatorics. Among the highlighted results was progress on the existence of non-sofic groups, a major open question in group theory, as well as work connected to Paul Erdős’s 1946 plane unit-distance problem and new circuit-complexity lower bounds for computing the matrix permanent. OpenAI said the mathematical arguments came from Astra, while human researchers organized the findings into papers and translated the proofs into Lean, a formal proof language that allows each step to be checked by software. That machine-verifiable layer is central to the claim: correctness can be audited computationally, even before conventional peer review. The company estimated token costs at roughly $2,000 when priced against its Sol model API rates, arguing that the expense per problem was modest and that additional compute could be applied. Several caveats remain. The Astra build was internal and unreleased, the results have not completed peer review, and no Millennium Prize problem was solved. The development also matters for teams designing an AI Crypto Wallet, because machine-verifiable reasoning could influence how autonomous agents are authenticated, audited and held accountable. Still, external mathematician Thomas Bloom described the work as significant, making the episode a concrete benchmark for AI research capability at a time when Worldcoin and other identity-focused networks are trying to separate genuine human participation from machine activity.

For Worldcoin and the wider Altcoin sector, the second AI industry disclosure touches product restraint rather than model capability. OpenAI product lead Tibo Sottiaux said on X on Aug. 1, 2026 that he had worked on a Google internal project called LMChat, which he described as essentially ChatGPT one year before ChatGPT launched. Sottiaux, who joined OpenAI in 2024 and later took responsibility for core products including ChatGPT and Codex, said Google was too nervous to release the system and that DeepMind faced internal limits preventing launches that might cannibalize Google’s search business. The timeline aligns with ChatGPT’s November 2022 debut and Google’s earlier LaMDA demonstration in 2021. The account also echoes earlier Google conversational work: Meena existed in 2020, and engineers Noam Shazeer and Daniel De Freitas left in 2021 to found Character.AI after the company hesitated to bring such products to market. Google chief scientist Jeff Dean has previously acknowledged that the company had internal chatbot technology but believed it was not better than direct search, while executives worried that erroneous answers could damage trust built over decades. The cost of hesitation became visible in February 2023, when Google’s rushed Bard launch included a demonstrable answer error and Alphabet’s market value fell by about $100 billion in a single session. Google later merged its Brain and DeepMind units to accelerate AI development. For Worldcoin (WLD), the episode is less about one product and more about how large platforms manage trust, safety and release cycles when AI interfaces meet billions of users. That context matters because WLD’s market narrative often tracks discussions around AI safety, identity and platform competition. As AI agents become more common, decisions around authentication, user consent and verification could shape demand for identity-oriented tokens and tools, including interfaces that might one day connect to an AI Trading Bot or a self-custodial service during a bear market.

COINOTAG’s proprietary 42-indicator composite S/R scoring engine frames Worldcoin (WLD) as attempting a recovery within a downtrend at $0.3212, up 4.80% over 24 hours as of publication. The nearest resistance at $0.3329 scores 61/100, driven by Fibo 0.786 confluence, while support at $0.2952 scores 58/100 from BB Lower and Keltner Lower. RSI at 38.64 and a bearish MACD signal leave momentum fragile. Derivatives show a negative perp funding rate of -0.0077% and $63.1 million open interest, suggesting short-side crowding rather than confident longs. With Fear and Greed at 27, a reclaim of $0.3329 could target the $0.4066 resistance; loss of $0.2952 would invalidate the bounce and expose weaker $0.2517 support.