AdvertiseFee Deal Desk

NEAR Protocol

NEAR Co-Founder Illia Polosukhin Warns of “Security Armageddon” Across the Internet

NEAR co-founder Illia Polosukhin warns of “Security Armageddon” across internet software, citing PHP, Node.js and LiteLLM flaws as NEAR slips 3.8%.

Be a creator
October 22, 2026, 08:18 AM UTC5 min read
AI SummaryAI
  • Illia Polosukhin warned of “Security Armageddon” across internet software on October 22.
  • Polosukhin named PHP, WordPress, malicious TLS certificates, Node.js and LiteLLM as weak points.
  • PHP maintainers disclosed a TLS certificate memory flaw in a September 24 advisory.
  • NEAR Intents resumed operations October 1 after a $3.8 million USDT exploit on BNB Smart Chain.
binance.com

The October 22 Warning

In a post on X published on October 22, NEAR Protocol co-founder Illia Polosukhin declared that the internet has reached “Security Armageddon,” arguing that a large share of the web is affected in some way today as older software vulnerabilities surface and get chained into further exploitation. He framed the risk as structural rather than isolated: weaknesses in long-lived components, authentication systems and shared infrastructure leave the applications above them exposed, and one flawed dependency can drag down every service that trusts it. His list of live pressure points is specific: PHP systems, WordPress among them, which still power much of the web’s content layer on components maintained for years; malicious TLS certificates; Node.js; and LiteLLM. Each sits at the base of stacks few operators ever inspect, which is why he treats their upkeep, not any single bug, as the systemic problem. The response he sketched is equally concrete. Uncensored AI models could help security teams examine defects and answer threats faster, he argued, while formal verification, a method that checks software against precisely defined mathematical requirements covering authorization and transaction handling, offers a second line of defense. A third proposal replaces conventional API keys and access tokens with cryptographic authentication, paired with hosting able to shield software even from the platform providers running it. For operators, the immediate instructions were ordinary: apply updates, patch the affected components and rotate credentials. The warning landed with the NEAR price near $5: our live reading shows a 3.8% drop over the past 24 hours, against an 8.21% gain over the prior week, a 13.21% monthly advance and a market capitalization around $6.57 billion. Skeptics may file the “Armageddon” label under FUD, but the founder’s case rests on published advisories, not sentiment.

The records behind the list are public. On September 24, PHP maintainers published a security advisory describing a memory-handling defect triggered by specially constructed TLS certificate names; the flaw touched specific versions of PHP’s OpenSSL integration and was fixed in later maintenance releases. The document demonstrates a possible memory-access failure, not a confirmed wave of attacks on WordPress sites. Node.js ran into separate trouble on October 15, when maintainers of Undici, its HTTP client library, disclosed vulnerabilities spanning certificate validation, connection handling and denial-of-service conditions. LiteLLM’s maintainers had earlier documented a flaw that could let authenticated users manipulate routing parameters and, through server-side requests, reach provider credentials. Read together, the advisories establish concrete weaknesses without proving any single coordinated campaign, which is the distance between Polosukhin’s rhetoric and the underlying documents. The remedy side carries its own caveat: he conceded that dependable AI systems able to translate software behavior into mathematical representations, then verify those representations against the original programs, do not yet exist at the needed standard. NEAR’s research line points the same way. Work published on October 19 described an AI agent solving 672 mathematical proof problems at an estimated model cost of $111.85, positioned as progress toward automated verification while explicitly distinguishing math benchmarks from proving deployed applications secure. The theme also has a live precedent on the network itself. NEAR Protocol (NEAR) Intents, the protocol’s cross-chain transaction infrastructure, resumed operations on October 1 after a $3.8 million exploit involving USDT on BNB Smart Chain. The project’s SHIELD security system flagged unusual activity, operators paused affected services, and developers corrected the vulnerability within an hour of detection, pledging full compensation while the base chain and unrelated applications stayed untouched. No link ties that incident to the flaws in the latest warning; both, though, illustrate what interconnected services risk. As our earlier report on NEAR Intents cross-chain swap volume topping $130 million in 24 hours showed, this component already carries serious traffic, which is why readers weighing exposure may start with our guide to blockchain security audits.

Futures Open Interest Falls 7.45%

The market’s first response reads as caution, not alarm. Derivatives data shows crypto futures open interest down 7.45% at $1.50 billion, with futures volume 6% lower at $2.56 billion; the retreat in leveraged margin trading positions is deleveraging, though open interest alone cannot say whether longs or shorts closed. Spot trading data recorded roughly $9.21 million in net outflows on October 22, thinning exchange supply without proving accumulation, while daily volume stood near $1.08 billion. On the chart, the token sits below its middle Bollinger Band at $5.066 with the Relative Strength Index at 52.88 against a 55.85 average; the support and resistance levels in play are $4.80 and $4.51 to the downside and $5.62 above. Analyst Rekt Capital maps a wider range of $3.79–$5.97, while Michaël van de Poppe holds that losing $4.30 would break the token’s higher-low structure. What stays unproven is the core of Polosukhin’s remedy: no production AI system yet performs the software-to-mathematics translation he calls for, so the steps available today remain the ones he gave operators, patch, rotate credentials and repeat.

COINOTAG's editorial and research desk.

AI-Assisted

AI-generated, AI-reviewed, under COINOTAG editorial oversight.