OpenAI
Oct 22, 2026

OpenAI model helps identify programming language flaw affecting millions of sites

An OpenAI model helped uncover a potentially catastrophic cybersecurity flaw in a programming language used to run millions of websites and applications, according to Techub, citing Bloomberg. The report did not name the language, but said the issue was serious enough to be described as potentially catastrophic. The disclosure was discussed on Bloomberg Tech by OpenAI President Greg Brockman and Cloudflare CEO Matthew Prince. In that conversation, the two executives talked about how AI is strengthening both sides of the cybersecurity equation: defenders trying to protect systems and attackers looking for new ways to exploit them. Cloudflare was also involved before the vulnerability became public. According to the report, the company helped protect websites ahead of the public disclosure. No additional technical details, timeline for the flaw’s discovery, or remediation specifics were provided in the source material.

770
OpenAI model helps identify programming language flaw affecting millions of sites
Solana
Oct 20, 2026

Asymmetric Research says Solana governance program svmgov had voting flaw fixed before key proposal vote

Blockchain security firm Asymmetric Research disclosed that Solana’s on-chain governance program, svmgov, previously contained a flaw that could have affected vote counting under a specific condition. If a validator did not participate in voting, independently submitted votes from its delegators made in advance might not have been included in the final tally. According to the disclosure, the issue was reported to the Solana Foundation on July 10 and patched on mainnet on Aug. 3, ahead of the Aug. 24 governance vote. Asymmetric Research estimated from on-chain data that, without the fix, support for proposal SGP-0002, which aimed to speed up the decline in SOL inflation, could have fallen from the actual 67% to 66.65%, below the 66.67% approval threshold. The flaw was fixed before voting took place and did not change the proposal’s final outcome.

1040
Asymmetric Research says Solana governance program svmgov had voting flaw fixed before key proposal vote
XRP
Oct 11, 2026

XRP Kuwait says coordinated node updates do not amount to centralized control

XRP Kuwait, an operator of a validator node, has pushed back against criticism that followed the rollout of the xrpld 3.4.1 emergency patch, arguing that near-simultaneous upgrades by validators should not be treated as evidence of centralized control on the XRP Ledger. The criticism emerged after nodes appeared to install the patch at almost the same time, prompting questions over whether the network retained meaningful validator independence. The patch addressed an integer overflow flaw in the payment engine that had existed since 2015. According to the report, the bug could let a sender lose only a very small amount while a recipient still received the full amount, effectively creating unauthorized XRP. A separate patch also closed a vulnerability tied to the Batch mechanism. Critics focused on the fact that a protection measure took effect immediately after the upgrade and was not approved through a network-wide vote, raising fresh questions about validator autonomy. The report cited U.Today as the source.

1040
XRP Kuwait says coordinated node updates do not amount to centralized control
XRP
Oct 10, 2026

XRP Ledger bug could have minted spendable XRP, RippleX says no public network abuse found

A security report released Friday said a flaw in the XRP Ledger payment system could, in theory, have allowed attackers to create large amounts of new XRP at almost no cost, breaking the token’s fixed-supply design. The issue was traced back to 2015, according to the report, and was discovered by researcher Cayden Liao and Veria AI, who privately disclosed it on Sept. 22. RippleX engineers later reproduced the exploit on an isolated server and confirmed that the newly created XRP could be spent in subsequent transactions. RippleX said it found no evidence that the bug had been exploited on any public network. The report said the attack path involved the ledger’s built-in exchange, where accounts post offers to swap one token for another. By spreading trades across hundreds of accounts, an attacker could allegedly trigger a counting error large enough for seller accounts to receive full XRP payouts while buyer accounts were barely debited. Developers shipped a fix in xrpld 3.4.1 on Sept. 25 without publicly detailing the issue at the time. The case adds to a recent run of long-hidden crypto vulnerabilities that have surfaced with AI-assisted research.

1210
XRP Ledger bug could have minted spendable XRP, RippleX says no public network abuse found
Core Lightning
Sep 21, 2026

Core Lightning urges node operators to disable experimental feature over channel fund risk

Core Lightning has urged node operators to immediately disable an experimental feature while developers investigate a vulnerability that could put channel funds at risk, according to a post from Bitcoin News on X. The warning marks the second alert from Core Lightning in recent weeks. Earlier, the project released an August patch and later rolled out version 26.06.7 after a series of AI-generated CVE reports. Core Lightning has not disclosed how the experimental feature could be exploited. The notice focuses on risk containment while the investigation is still underway, with no technical exploitation details released so far.

980
Core Lightning urges node operators to disable experimental feature over channel fund risk
Balancer
Sep 4, 2026

Balancer Sends On-Chain Message to V1 Attacker: Respond by Sept 9 or Face Escalation

Balancer has sent an on-chain message to the known address involved in the V1 attack, offering a channel for communication and fund return. The attacker must respond privately via Blockscan before 5:00 UTC on September 9. If funds are returned to the DAO multisig address, Balancer will negotiate a bounty and will not pursue legal action solely for the return. Failure to respond will lead to technical, on-chain, and legal measures, with the bounty redirected to whistleblowers who help identify the attacker. This follows Balancer's August 31 disclosure of a V1 contract vulnerability that allowed LP funds to be drained.

1760
Balancer Sends On-Chain Message to V1 Attacker: Respond by Sept 9 or Face Escalation
Total 41 items
Go topage