AI Worm Adapts Attacks in Real Time, Anthropic Ships Mythos 5 as Altcoin Index Stalls at 49

Crypto News

A team of academic and corporate researchers has demonstrated a proof-of-concept AI-powered worm capable of finding vulnerabilities, devising tailored attack paths, and replicating across networks without any human operator. Unlike fixed-exploit malware such as WannaCry, the system uses a large language model to reason about each target and synthesize attack logic on the fly. Critically, it runs locally on infected machines using open-weight models rather than relying on cloud services, making it harder to disrupt. In an isolated test network of 33 Linux, Windows, and IoT systems seeded with common flaws, the worm adapted its tactics across 15 separate experiments, signaling that autonomous generative adversaries have moved well beyond theory.

Market commentators argue there will be no broad altcoin season until aggressive monetary easing returns. The Altcoin Season Index sits at 49, a reading that confirms Bitcoin dominance rather than a rotation into smaller tokens. Bitcoin itself has slid from roughly $77,000 in mid-May toward $61,282, a decline of about 24% over 30 days. Analysts describe altcoins as trophy assets that only attract inflows when investors hold cash they can afford to lose, meaning they bleed hardest when liquidity tightens. The view points to 2028 or 2029 as a more realistic window for any market-wide altcoin breakout.

Anthropic has officially launched Claude Mythos 5, a restricted-access model the company says carries unprecedented cybersecurity capabilities, alongside Claude Fable 5 for broader public use. Mythos 5 will initially be available only to approved cybersecurity organizations, critical infrastructure operators, government partners, and selected life sciences researchers, with the firm warning that Mythos-class systems present significant risks. Fable 5 routes some sensitive requests to a less capable model. The release follows months of scrutiny after a preview version reportedly completed a 32-step corporate network intrusion exercise without human help and surfaced more than 271 vulnerabilities in the Firefox browser.

The worm research underscores a sharpening threat to blockchain infrastructure, exchanges, and self-custody setups. Because the malware generates fresh attack strategies per target, traditional defenses built around patching known vulnerabilities offer limited protection. Researchers warn that self-propagating attacks have already shown real-world reach, citing the Shai-Hulud malware that spread through software used by major AI firms. For crypto operators, the implication is that wallet infrastructure, node software, and custody systems may face adversaries that probe and adapt continuously, raising the premium on hardware isolation, cold storage, and rigorous operational security across the entire transaction stack.

The structural case against altcoins extends beyond monetary policy. Dashboard data shows 256 days have passed since altcoins last led the market, while Bitcoin recorded its most recent dominant stretch only 21 days earlier. Historically, altcoin seasons are brief and infrequent, averaging around 17 days and arriving roughly every 67 days. Supply dynamics compound the difficulty: where the market held about 3,000 tokens in 2017, it now counts tens of millions, scattering capital across far more tickers. That dilution means any single asset captures a thinner slice of inflows, reinforcing a bear market structure for the long tail of digital assets.

Anthropic frames the dual release as a deliberate safety experiment. The company began Project Glasswing in April, releasing an earlier Mythos-class model to a limited group of cyber defenders and critical software providers while it evaluated misuse risks. It now says it will expand Mythos access through a broader trusted-access program for cybersecurity work and biomedical research, conditioned on safeguards strong enough to reliably prevent abuse. The tension is explicit: the same capabilities that let defenders discover hundreds of vulnerabilities could empower attackers, placing frontier AI labs at the center of a defensive-offensive balance that increasingly overlaps with digital-asset security.

Taken together, these developments sketch a cycle defined less by speculative exuberance than by consolidation and risk. Capital is concentrating in Bitcoin while the broader circulating supply of alternative tokens dilutes returns, and the dominant narrative has shifted from retail euphoria toward institutional caution. Layered on top is an escalating security dimension, as adaptive AI malware and unprecedented model capabilities reshape the threat landscape for exchanges, custodians, and protocols alike. The through-line is a maturing market navigating tighter liquidity and a more dangerous technological frontier, where survival favors disciplined infrastructure over momentum-driven bets.