The Hacker News | #1 Trusted Source for Cybersecurity News — Index Page

Autonomous security agents are getting good at finding bugs. Nobody has a good way to measure how good. Point one at a realistic target and what comes back is a report the agent wrote about itself: confident prose, a list of findings, and no way to tell which of them happened. Someone with a security background then sits down and checks every claim against the target. Which findings are real, which are duplicates, which are inventions, and, the question nobody has time for, what did the agent never try? That is a day of expert work for one run. Multiply it by three models, four prompt variants and ten repetitions, and the review queue is longer than the experiment. XRanges for AI, built by CTF.ae, exists for that loop. It deploys realistic target applications with instrumentation baked into every service, records what an agent actually does inside them, and scores each run live on four independent signals. This walkthrough covers how it works, what a run looks like from deployment t...

Anthropic and OpenAI on Tuesday announced new models, with both artificial intelligence (AI) companies noting that they are continuing to invest in improving alignment to combat risky behavior. Opus 5.5, per Anthropic , is a "major step up from Opus 5," and "achieves the best scores of any model to date on our automated behavioral audit, our alignment suite that tests Claude across thousands of simulated scenarios." The AI company said the model is less likely than its other recent models to carry out hard-to-reverse actions or act outside the boundaries it's been given, adding it's more resistant than Opus 5 to prompt injection. In its systems card, Anthropic explained that the latest model exhibited less misaligned behavior and less cooperation with misuse than any other recent Claude model. "It also took overeager or destructive actions less than any other model we tested," Anthropic said. "Regressions included being more likely ...

A use-after-free in the Linux kernel's AF_UNIX socket subsystem can be used to escape a container and gain root on the host, security firm DepthFirst  said in research published September 22. The flaw, tracked as  CVE-2026-80521  (CVSS score: 7.8), was fixed upstream on August 6, but Ubuntu has not shipped the patch for its 26.04, 24.04, or 22.04 LTS releases. DepthFirst  released exploit code  targeting Ubuntu 26.04. Ubuntu's security tracker  lists the Linux package on 26.04 as "vulnerable, work in progress." The 24.04 and 22.04 releases are also affected through newer kernel packages, including those for AWS, Azure, and GCP workloads. No fix has shipped on any affected release. The flaw is not in CISA's Known Exploited Vulnerabilities catalog, and there are no confirmed reports of attacks using it. The vulnerability sits in the kernel's garbage collector for AF_UNIX sockets. That collector cleans up file descriptors passed between processes t...

Attackers are exploiting a critical flaw in F5 BIG-IP Access Policy Manager (APM) that lets them run code on a BIG-IP system without logging in, F5 says. The flaw,  CVE-2026-94127 , affects only systems in which APM serves as an OAuth authorization server, issuing access tokens to applications. F5 disclosed it in  an advisory  on September 22 and has released engineering hotfixes. APM is the BIG-IP module that controls how users reach an organization's applications and networks. The vulnerable setup has an APM access policy and an OAuth authorization server profile on the same virtual server, which hosts the BIG-IP address that receives the OAuth traffic. Specific malicious traffic sent to that virtual server can lead to remote code execution. The flaw is a heap-based buffer overflow. F5 rates it 9.8 out of 10 on CVSS v3.1 and 9.3 on CVSS v4.0. Because the malicious traffic goes to the virtual server itself, limiting access to the BIG-IP management interface does...

A Chinese threat actor codenamed UTA0565 has been observed exploiting the recently disclosed Google Chrome-Microsoft Windows exploit chain as zero-days through fake websites. The attacks, detected on September 3 and 4, 2026, involved the chaining of two vulnerabilities in Chrome (CVE-2026-85046, CVE-2026-87491) and one impacting Windows Advanced Local Procedure Call (CVE-2026-85880) to break out of the browser's sandbox and achieve remote code execution. "UTA0565 masqueraded as various entities including media organizations and a non-governmental organization (NGO)," Volexity researchers Damien Cash and Tom Lancaster said in an analysis published this week. "Notably, this threat actor's campaigns differed from previously documented attacks by using multiple fake websites to deceive victims." One such campaign targeted Asian government entities with Chinese- and English-language phishing emails that urged recipients to support Hong Kong activist Cho...

A new security vulnerability in Next.js could allow attackers to run code on a server via  ImageResponse , the feature that generates Open Graph and other social preview images, Vercel said. The risk applies when an app puts values an attacker controls, such as text read from the request URL, into the image. Vercel, which develops Next.js,  fixed the flaw  on September 22 in version 16.3.6. The flaw, tracked as CVE-2026-94545 , affects Next.js 16.2.0 through 16.3.5 when ImageResponse runs on the Node.js runtime, which Next.js  uses by default . Vercel's  advisory  rates it critical, with a CVSS score of 9.5. The Edge version of ImageResponse is not affected, and neither is Next.js 15. ImageResponse uses  Satori , a Vercel library, to convert the image layout into SVG code before the final PNG is generated. Affected apps are those that "pass attacker-controlled values into SVG content, attributes, or styles during image generation", according to ...

The cyber extortion group known as ShinyHunters on Tuesday claimed it had breached the U.S. Federal Bureau of Investigation and stolen data belonging to current and former employees at the agency. "We have compromised the FBI. We hold very sensitive data on almost ALL FBI Agents and individuals who filed an application with the FBI for a job," the group said in a statement posted on their dark web site. "Whether it be a Special Agent or any other role within your agency. The following FBI services were compromised: Criminal Justice (CJ), HR, Medlink, and more." The development was first reported by 404 Media. ShinyHunters said the FBI was targeted in response to a May 2026 public service announcement (PSA) that detailed the threat actor's targeting of Canvas , an online Learning Management System (LMS), while urging victims not to pay. The attackers, in their own counter PSA, described them as "substantial false allegations," adding, "w...

Attackers exploited a previously unknown flaw in Check Point's Security Management Server in a handful of targeted attacks on July 23,  the company said . The flaw, CVE-2026-93616 , allows an attacker who can access the server's web service to run scripts on it without logging in. Check Point released a fix on September 22 for the server that controls firewall policies for the Check Point gateways it manages. Separately,  Check Point said  attackers have been trying since September 12 to exploit a VPN flaw it  fixed on September 9 . The attempts, against a flaw tracked as CVE-2026-85102, have targeted customers of Spark, Check Point's firewall line for small businesses. When the fix came out, Check Point had no evidence the flaw was being exploited. CVE-2026-93616 is a path traversal bug in the management server's web service. The service does not properly limit which files and folders a request can reach. An attacker can use it to upload scripts to the s...

WordPress has fixed a critical flaw in its core software that lets an attacker with no account make a site load a PHP file from outside its theme folders. On some servers, that can go further, allowing the attacker to run their own code. The fix shipped on September 22 in WordPress 7.1.2, with fixes for every branch the project still supports, back to 4.7, and WordPress is telling site owners to  update now . WordPress rates the flaw as critical, assigns it a CVSS score of 9.2, and assigns it  CVE-2026-87902 . Reaching it requires no account and no action from a logged-in user. Every version from 4.7.0 through 7.1.1 is affected. That includes 7.1.1, from WordPress's  September 17 security release , so a site updated less than a week ago still needs this one. It is a separate flaw from the ones that the release fixed.

Cybersecurity researchers have disclosed details of a malicious npm package named "tw-pkgprobe-7731" that masquerades as a security tool targeting developers integrating Twilio into their applications, while stealthily attempting to harvest sensitive data. The package, named "tw-pkgprobe-7731," was first uploaded to the npm registry in mid-August 2026 by an npm account named "twdepprobe7731." In total, 11 versions of the package were published in quick succession on the same day over an approximately 45-minute time period. The npm user account no longer exists as of writing. "The first version of tw-pkgprobe-7731 posed as an authorized security research probe," ReversingLabs researcher Lucija Valentić said in a report published today. "Comments inside the package describe it as an 'Authorized bug-bounty research probe (Twilio HackerOne program)' that 'runs only inside Twilio's serverless packager sandbox' and ...

Microsoft on Tuesday announced the takedown of the EvilTokens device code phishing service that it said used artificial intelligence (AI) "at every step of the attack chain." The action, carried out with authorization from the U.S. District Court for the Eastern District of Virginia, involved the efforts of Health-ISAC, alongside Cloudflare, Coinbase, OpenAI, Railway, SpyCloud, The Shadowserver Foundation, and TRM Labs. Microsoft is tracking the threat actors behind the development and support of EvilTokens as Storm-2992 . In tandem, the Metropolitan Police Service arrested two men, aged 32 and 38, on September 11, 2026, in connection with the illicit commercial operation. The tech giant described EvilTokens as a "powerful cybercrime platform" that used AI to compromise email accounts and design roadmaps for financial fraud and scams. "While EvilTokens helped cybercriminals access email accounts, at the center of the service was an AI-style chatbot th...

A critical vulnerability in  Bifrost , an open-source AI gateway that routes requests to more than 20 LLM providers, allows an unauthenticated attacker to run arbitrary commands on the gateway server with a single HTTP request. The flaw, tracked as  CVE-2026-90898  (CVSS score: 9.8), affects all versions of the Bifrost HTTP transport before 2.1.0 when management authentication is disabled, which is the default configuration. A fix is available in transports/v2.1.0. Yuval Moravchick of  JFrog Security Research , who discovered the flaw, said an attacker can register a stdio-type MCP client through a single unauthenticated POST to the management API endpoint /api/mcp/client. Bifrost starts the specified command immediately, before any MCP handshake, as the gateway process user. On the official Docker image, that user is appuser. Because the gateway stores API keys for every connected provider, executing commands on the gateway process grants the attack...

A zero-day proof-of-concept tool that stops Microsoft Defender from installing platform and signature updates by filling all available disk space was  published on GitHub  on September 19. The tool, called BigDiskBuster , has no patch, no CVE, and no Microsoft advisory. Its author, Abdelhamid Naceri, is a former Microsoft security researcher whose earlier Defender exploits were used in attacks. When updates are blocked, Defender keeps running, but its detection content grows stale. The researcher's screenshot shows Defender returning a generic Windows error when trying to update, but whether the failure raises an automatic alert is not clear from the proof-of-concept alone. Naceri said he was dismissed from Microsoft's Security Response Center in 2024 and has been releasing exploits without coordinating with the company since April. His first three Defender tools — BlueHammer, RedSun, and UnDefend — were all  exploited in live intrusions  before Microsoft...

Security teams have spent decades asking whether an identity has too much access. AI agents raise a harder question: how can we determine which paths an autonomous system can discover, given the access it already has? A person may try several ways to complete a task. A deterministic application follows the flow its developer wrote. But an AI agent is relentless in its pursuit of done. In May 2026, OpenAI announced that one of its models had disproved a 1946 Erdős conjecture in discrete geometry, largely by working through paths a mathematician would abandon as too tedious. The same principle applies to cybersecurity. An AI agent can test thousands of actions, abandon failed routes, discover credentials, switch tools, and keep going. That persistence is part of what makes agents useful, but it also changes how we need to think about lateral movement. AI agent risk has two dimensions: Access defines the possible blast radius Autonomy determines how much an agent can do with...

Attackers are exploiting a new flaw in on-premises VeloCloud Orchestrator (VCO), the server that manages the Edge devices in a VeloCloud SD-WAN, Arista said on September 22. The flaw, tracked as CVE-2026-93952 , may allow a remote attacker with no login access to privilege internal functions and affect the VCO host. Only orchestrators set up to authenticate their Edges with certificates are exposed. As of September 22,  fixed releases  are out for the 5.2 and 6.4 release trains, but not yet for the 6.1 and 7.0 trains. Arista has already patched the Hosted and Dedicated versions of VCO. The affected releases include those that fixed a different VCO flaw, which Arista reported as  exploited in July . Arista gave the flaw a CVSS 3.1 score of 10.0. A successful attack may compromise the orchestrator and the data it manages. A compromised VCO may also give attackers access to the Edge devices it manages. Arista said the flaw "was discovered externally and is known t...

When the Digital Operational Resilience Act (DORA) became enforceable across the European Union in January 2025, it triggered an administrative sprint. Financial entities spent the first year establishing risk governance, assessing third-party service providers, updating contract clauses, and documenting incident escalation workflows. Now in its second year, the harder part of DORA is demonstrating how well frameworks work in practice. EU regulators are increasing their focus on DORA implementation, Information and Communication Technology (ICT) incident analysis, and the effectiveness of ICT risk supervision. For security teams, that raises an important question: does the SOC have enough visibility to detect, investigate, and scope an active intrusion across critical systems? While DORA doesn’t prescribe a particular security stack, several of its requirements rely on continuous visibility in the ICT environment to identify behavior that may indicate an emerging risk. Continuou...