Top stories.
- Governments and industry race to harness AI for vulnerability discovery.
- FIRESTARTER malware remained on Cisco devices after patches were applied.
- Cloud development platform Vercel confirms breach.
- Surveillance vendors exploit telecom weaknesses.
- Sean Plankey withdraws nomination to serve as CISA director.
Governments and industry race to harness AI for vulnerability discovery.
Axios reports that the US National Security Agency (NSA) is using Anthropic's Mythos Preview model, despite the Pentagon labeling the AI company as a supply chain risk. Mythos excels at finding software vulnerabilities, which makes it valuable for both defensive and offensive operations. Anthropic has granted around 40 organizations access to the model through Project Glasswing, which aims to apply Mythos defensively with industry partners to secure critical infrastructure. Neither Anthropic nor the NSA has commented on the agency's reported use of the tool. Axios notes that MI5, the NSA's UK counterpart, said it has access to Mythos through the AI Security Institute.
Anthropic is embroiled in a lawsuit over its designation as a supply-chain risk, although the company's CEO Dario Amodei met with senior White House officials last week in a meeting that both sides described as "productive." The Pentagon labeled Anthropic a risk after the company insisted on limitations to surveillance- and weapons-related development during contract negotiations earlier this year. The NSA's reported use of Mythos highlights a tension between the practical value of advanced AI capabilities and the government's public dispute with Anthropic.
Mozilla, one of the companies granted access to Mythos, said on Tuesday that its latest Firefox update contains fixes for 271 vulnerabilities identified using the AI tool. Mozilla says the model is "every bit as capable" as an elite human researcher, noting that all of the vulnerabilities could have been discovered by humans given a longer timeline. The company takes an optimistic view of AI-assisted vulnerability discovery, stating, "A gap between machine-discoverable and human-discoverable bugs favors the attacker, who can concentrate many months of costly human effort to find a single bug. Closing this gap erodes the attacker’s long-term advantage by making all discoveries cheap."
At Black Hat Asia in Singapore, RunSybil CEO Ari Herbert-Voss said open-source AI models can identify software vulnerabilities as effectively as Anthropic's restricted Mythos model when used together in coordinated workflows, the Register reports. Herbert-Voss attributes Mythos's strength to "supralinear scaling," where allocating more training resources produces disproportionately greater results. He added that open-source models can achieve similar results when several of them are run together in a "scaffolding."
Meanwhile, SecurityWeek reports that one of China's largest security firms, Qihoo 360, claims that its cybersecurity-focused AI model discovered more than a thousand vulnerabilities during the Tianfu Cup hacking competition. ETH Zurich researcher Eugenio Benincasa analyzed this claim and concluded that 360's model is approaching Mythos's reasoning capabilities, but hasn't drawn even yet. SecurityWeek notes that Antropic's own CEO has estimated that open-source models and Chinese companies could match Mythos's cybersecurity capabilities within six to twelve months.
FIRESTARTER malware remained on Cisco devices after patches were applied.
A state-sponsored APT deployed a backdoor on Cisco security devices running ASA or Firepower, exploiting two vulnerabilities (CVE-2025-20333 and CVE-2025-20362) that were patched in September, CyberScoop reports. Notably, the malware survives patches by embedding a persistence mechanism in the device's boot sequence, and devices that were breached before patches were applied may still be compromised. Cisco says customers can mitigate the infection by reimaging their devices.
The US Cybersecurity and Infrastructure Security Agency (CISA) and the UK's National Cyber Security Centre (NCSC) published a joint report on Thursday analyzing the backdoor, dubbed "FIRESTARTER," which was discovered within the network of a US Federal agency. The report states, "FIRESTARTER is a Linux Executable and Linkable File (ELF) designed to execute on Cisco Firepower and Secure Firewall devices, serving as a C2 channel for remote access and control. The malware achieves persistence by detecting termination signals and relaunching itself, and it can survive firmware updates and device reboots unless a hard power cycle occurs."
While the security agencies and Cisco don't attribute the campaign to a particular nation-state, Cisco links the operation to an earlier attack campaign dubbed "ArcaneDoor," and CyberScoop notes that Censys researchers tied the ArcaneDoor campaign to a China-based threat actor.
Cloud development platform Vercel confirms breach.
Cloud development platform provider Vercel has confirmed a breach after a hacker claimed to have stolen data from the company. Vercel, which developed and maintains the Next.js React framework, said in a statement on Sunday, "Initially we identified a limited subset of customers whose Vercel credentials were compromised. We reached out to that subset and recommended an immediate rotation of credentials. If you have not been contacted, we do not have reason to believe that your Vercel credentials or personal data have been compromised at this time."
The company said the incident "originated from a small, third-party AI tool whose Google Workspace OAuth app was the subject of a broader compromise, potentially affecting its hundreds of users across many organizations." Vercel's CEO, Guillermo Rauch, added in an X post that the third-party AI provider was Context.ai.
A threat actor using the handle "ShinyHunters" posted on an underground forum offering to sell "multiple employee accounts with access to several internal deployments, API keys (including some NPM tokens and some GitHub tokens)." BleepingComputer notes that the ShinyHunters extortion group said it wasn't involved in the breach and is not affiliated with this user.
Surveillance vendors exploit telecom weaknesses.
Two separate surveillance operations are abusing known weaknesses in telecom signaling infrastructure to track users' locations, TechCrunch reports. Researchers at the University of Toronto's Citizen Lab discovered the campaigns, explaining that the attacks exploit vulnerabilities in SS7, an older protocol that lacks basic security mechanisms, and in Diameter, a newer protocol often deployed without full protections.
The surveillance campaigns were observed targeting high-profile individuals, including a well-known company executive in the Middle East. Citizen Lab says the "single user targeting across multiple operator networks and country jurisdictions persisting for years is characteristic of a commercial surveillance platform deployed across multiple operators and likely used by multiple global clients."
The operations targeted telecoms based in at least seventeen countries across Europe, Africa, and Asia. Citizen Lab states, "This pattern indicates a centralized surveillance C2 platform with deep integration into the signalling ecosystem, providing multiple routing options to covertly reach target networks around the world." The researchers don't attribute the operations to any particular surveillance vendors, but note that the attacks used infrastructure based in Israel, the United Kingdom, and the Channel Islands.
Sean Plankey withdraws nomination to serve as CISA director.
Sean Plankey, President Trump's choice to lead the US Cybersecurity and Infrastructure Security Agency (CISA), withdrew his nomination for the post on Wednesday, POLITICO reports. Plankey had awaited confirmation for over a year as the GOP Senate majority failed to obtain the votes needed to advance his nomination. Senator Rick Scott (Republican of Florida) worked to prevent Plankey's confirmation due to a disputed Coast Guard shipbuilding project.
Plankey sent a letter to the White House stating, "After thirteen months since my initial nomination, it has become clear the Senate will not confirm me. While I humbly request the removal of my nomination, I wholeheartedly support President Trump’s upcoming nomination for CISA and look forward to the continued success of the United States of America."
