Archive for July, 2026

GitHub and PyPI both just closed the gap attackers rely on for supply chain attack

Posted in Commentary with tags , on July 27, 2026 by itnerd

GitHub and PyPI both introduced new policies this week aimed at slowing down how fast malicious code can spread through the software supply chain. GitHub’s Dependabot will now wait at least three days before opening a pull request for a new package release, and PyPI is blocking new file uploads to any release older than 14 days, both closing the same window attackers have relied on: the gap between a malicious release going out and anyone noticing.

SecurityWeek has more:

To prevent the fast delivery of malicious code through the immediate fetching of brand-new releases, GitHub has introduced a Dependabot cooldown, where the automation tool waits for at least three days after a release has been published before opening a pull request.

“Waiting a few days before adopting a new release gives maintainers, security researchers, and automated scanners time to spot a malicious version and get it pulled before it ever reaches your pull requests,” GitHub explains.

The three-day cooldown only applies to non-security version bumps, and the behavior can be modified through the configuration option in the dependabot.yml.

“Three days as the default balances two goals: it pushes you past the window where most of these attacks live, and it doesn’t hold your dependencies back longer than necessary,” GitHub notes.

PyPI, on the other hand, is preventing the poisoning of releases older than 14 days by blocking the upload of new files to them.

“This restriction was put in place to prevent old and long-stable releases from being poisoned in case publishing tokens or workflows of PyPI projects were compromised. As far as we are aware, this has not yet been abused, but there is no technical reason beyond that attackers weren’t aware it was possible,” PyPI says.

The behavior will be enforced once ‘Upload 2.0 API’ and ‘Staged Previews’ have been standardized by PEP 694 and will affect only a small fraction of projects that still publish new files to older releases.

Maggie McDaniel, Global Head of Intelligence, iCOUNTER had this to say:

“This is exactly the kind of outcome effective public-private intelligence sharing is supposed to produce. Security researchers and the community kept surfacing the same pattern, a maintainer account gets compromised, a malicious release goes out, and automated tooling propagates it into thousands of projects before anyone has a chance to catch it, and instead of that staying a research finding, it became a platform-level policy change. A three-day cooldown on Dependabot and a 14-day lock on new uploads to existing PyPI releases won’t stop every attack, but they close the exact window that made this class of attack so effective: the gap between publication and detection. What I’d want to know now is whether other package ecosystems are watching this and building the same kind of delay into their own automation, because the ecosystems that don’t will be the ones attackers target next.”

This is a positive move. But it needs to be backed by knowing what goes into your applications without fail. Otherwise we will have a different version of the same problem.

Check Point zero-day highlights identity control shortfalls

Posted in Commentary with tags on July 27, 2026 by itnerd

With Check Point warning that attackers are actively exploiting the critical SmartConsole authentication bypass flaw (CVE-2026-16232) to obtain full administrative privileges, this is another reminder that security infrastructure itself has become a prime target. 

The vulnerability has already been exploited in the wild, prompting emergency patch guidance and inclusion in CISA’s Known Exploited Vulnerabilities catalog which can be found here..

Bojan Simic, CEO and co-founder, HYPR had this to say:

“Authentication bypass flaws like this happen when systems treat the token as the proof of identity instead of verifying what actually produced it. If an attacker can trick the application into issuing a token without a legitimate authentication event, that token becomes nothing more than a bearer credential. Whoever possesses it gets access, regardless of how it was obtained.

Passkeys fundamentally change that model. The private key never leaves the user’s device, so there is no shared secret, password hash, or reusable credential traveling across the network for an attacker to steal, replay, or manipulate into creating a valid session. By eliminating passwords and other static credentials, phishing-resistant authentication dramatically reduces the attack surface these flaws depend on.

However, authentication can’t stop at verifying the device. Organizations also need to verify the person behind it, particularly when granting privileged access or stepping up an administrative session. Device-bound cryptography establishes trust in the device; identity verification at high-risk moments establishes trust in the human. You need both to close the gap between someone possessing the right device and someone actually authorized to use it.”

It’s once again time to patch all the things. But this time the urgency is clear. Though it would make life a whole lot easer if flaws like this didn’t exist.

75 million Revolut records allegedly for sale on cybercrime forum

Posted in Commentary with tags on July 27, 2026 by itnerd

A threat actor has just listed a database allegedly containing 75 million records of Revolut customers on a cybercrime forum. The threat actor posted over 100 sample records. 

Cybernews researchers have investigated the dataset. Here’s what was included:

  • Partial credit card data: last 4 card digits, card type, expiration dates, and card statuses (active, blocked, frozen, etc).
  • User credentials: bcrypt or argon2id hashes, timestamps when the credentials were rotated.
  • User personal data: emails, full names, phone numbers, resident countries, addresses, currency, registration IP address, and other account identifiers
  • Other sensitive data: including device models, operational systems, and timestamps.

Cybernews has reached out to Revolut for comment. The company’s spokesperson said they are aware of the claims. However, they “see no indications of any breach.”

Cybernews researchers could not verify the claimed number of records and remain skeptical about the claimed scope of the dataset. If the dataset really contains 75 million records, the $500 price tag advertised on the illicit marketplace is unusually low for a leak of that size.

If the claims prove to be legitimate, the exposed information could pose serious risks. “Affected individuals may be at risk of social engineering attacks and identity profiling. Attackers may also attempt to gather full payment method information,” Cybernews researchers warned.

For more information, here’s the full report: https://cybernews.com/security/revolut-data-breach-75-million-records/

Inside DEF CON: New documentary offers first look behind closed doors in 13 years

Posted in Commentary with tags on July 27, 2026 by itnerd

For the first time since 2013, a documentary crew got access inside DEF CON, the world’s largest and most secretive hacker conference held annually in Las Vegas. The film features exclusive interviews with senior DEF CON figures, Black Badge winners, and even DARPA officials. The documentary will premiere on Cybernews’ YouTube channel on August 3rd.

This film explores how the past decade has changed not only public perception of hacker culture, but the community itself. As the line between hacking and defence has blurred, DEF CON now attracts everyone from teenagers to cybersecurity professionals to government agencies. 

Using personal archives from the community’s original creators and exclusive behind-the-scenes access, the documentary looks back at DEF CON’s early days and shows how far the community has come.

A promise to tell the truth

DEF CON has strict filming rules, and most previous coverage has rarely gone beyond vlogger-style footage. The Cybernews crew got exclusive sit-down interviews and tells the story of how the conference has changed from a small gathering into an event that now includes even government-organized competitions, including efforts to hack Air Force satellites.

‘Hacker’ through the eyes of the community

The community has long been followed by the myth that hackers are criminals. According to Moss, that perception dates back to the early 1990s, when reporters were trying to explain a new type of crime involving computers.

More than a decade of change: from parties to the US Air Force

Looking at DEF CON’s history, what used to be an old-school, underground hacker gathering with pranks and partying has evolved into something far broader. Today, it attracts professional coders, young people building careers in cybersecurity, and government officials. DARPA has held competitions at the conference, including the Cyber Grand Challenge and the more recent AI Cyber Challenge. 

When the US Air Force wanted more people to develop the skills needed to evaluate satellite security, it created the Hack-a-Sat competition at DEF CON and, in its final year, launched a satellite just for competing teams to attack.

The documentary will be publicly available on Cybernews’ YouTube channel starting August 3.

A-list directors, actors, and celebrities exposed in major film festival data breach

Posted in Commentary with tags on July 26, 2026 by itnerd

Cybersecurity researcher Jeremiah Fowler recently discovered and reported to ExpressVPN about multiple unsecured databases containing records linked to the Tribeca Film Festival. The exposed data included, but was not limited to:

  • 666,369 total records across three unsecured databases labeled “development,” “staging,” and “production,” spanning records from 2019 to 2026.
  • A production backup file with 295,916 records containing sensitive PII — email addresses, phone numbers, IP addresses, and hashed passwords — tied to users, film contacts, and more.
  • Confidential documents, press kits, and marketing materials, some explicitly marked as confidential, potentially exposing A-list talent and industry professionals associated with the festival.

Jeremiah’s full findings and key takeaways are published on this blog: https://www.expressvpn.com/blog/tribeca-film-festival-data-exposed/

AI agent claims to have found 19 Redis zero-days and built a working exploit in 27 minutes 

Posted in Commentary with tags on July 24, 2026 by itnerd

Researcher Chaofan Shou claims that Moonshot AI’s Kimi K3 agents autonomously found 19 Redis zero-day vulnerabilities in about 90 minutes, then built a working remote-code-execution exploit for one of them in 27 minutes. 

That’s charming. Actually it isn’t. I’ll get to that in a moment. Now I am going to get to some commentary by Arti Raman, CEO & Founder, Portal26

“Whether the specific numbers in this claim hold up under scrutiny or not almost doesn’t matter. What matters is that the capability being described, AI agents autonomously chaining vulnerability discovery into a working exploit in under half an hour, is no longer hypothetical. The question every enterprise running AI agents internally should be asking isn’t ‘could this happen to us,’ it’s ‘would we even know if it did.’ Most organizations have no visibility into what their own AI agents are actually doing with the access and tooling they’ve been given, which means an agent operating outside its intended scope wouldn’t look like an attack, it would just look like normal activity in a system nobody’s watching closely enough. You cannot govern what you cannot see, and right now most enterprises can’t see their AI agents at all.”

Roman Sannikov, Global Research Coordinator, iCOUNTER follows with this:

“The verified part of this story is notable enough on its own: two new Redis vulnerability classes, both patched, both capable of chaining memory corruption into full remote code execution. The unverified part, that a set of AI agents found 19 of these in 90 minutes and built a working exploit in 27, is the part worth treating carefully. Redis confirmed the flaws and the fixes. Nobody has independently verified the count, the timing, or how much of this actually happened without a human steering it.

That said, I wouldn’t dismiss it. We’ve been tracking a real trend of AI compressing the gap between a patch landing and a working exploit existing, and if even a fraction of this claim holds up, it’s consistent with that trend, not a departure from it. The takeaway defenders should draw from this isn’t ‘AI found 19 zero-days in 90 minutes,’ it’s that tools capable of something close to that now exist and are being tested in public. Whether this specific run is accurate or exaggerated, the capability itself isn’t hypothetical anymore, and threat intelligence teams should treat

At this point, I would assume that your opposition is using AI to attack you. Therefore you need to make sure that your defenses take that into account or you will be pwned.

UPDATE: Jacob Krell, Sr. Director: Secure AI Solutions & Cybersecurity, Suzu Labs (https://www.linkedin.com/in/jacob-krell) had this to say

“Intel’s Prescott chip hit 3.8 GHz in 2004 and Intel killed it because raw clock speed had become the wrong metric. K3’s 2.8 trillion parameter headline is the AI equivalent.

“The race now is efficiency, which K3’s own architecture proves with a sparse Mixture-of-Experts activating only 50 billion parameters per token at 2.5x K2’s scaling efficiency. Moonshot can give the weights away because you need 64 accelerators to serve them, and the moat is the deployment stack, not the weights themselves. The AI race will likely go to whoever puts GPT-4-class capability on consumer GPUs, not whoever adds another trillion parameters to their leading frontier model.”

Interestingly, Jacob explored this broader trend before this week’s news in a recent blog, arguing that AI’s next leap wouldn’t be smarter models – it would be AI becoming efficient enough to solve practical problems at scale. The Redis findings are an early proof point of that thesis. 

You can read it here. https://suzulabs.com/suzu-labs-blog/the-ai-industrys-prescott-moment

DentaQuest Starts Notifying 15 Million+ Individuals About May 2026 Cyber Incident 

Posted in Commentary with tags on July 24, 2026 by itnerd

The dental benefits administrator DentaQuest has started issuing notification letters to individuals affected by a May 2026 cybersecurity incident. The number of affected individuals has yet to be confirmed, although DentaQuest has confirmed that at least 15 million individuals have been affected.

Commenting on this is Paul Bischoff, Consumer Privacy Advocate at Comparitech: 

“This is a major data breach both in terms of the number of people affected and the types of personal information involved. DentaQuest customers should take advantage of free credit monitoring offered by DentaQuest and monitor their credit reports, bank accounts, and medical bills for unrecognized activity. Whether or not DentaQuest paid ShinyHunters’ ransom demand, there is no guarantee that the group will delete the stolen data. Breach victims should assume the worst and act accordingly to protect their accounts and identities.”

Comparitech recently published an in-depth research study looking at ransomware attacks against healthcare institutions in the first half of 2026, finding that attacks increased nearly 14% since the last half of 2025. 

WP Botnet Master – How a Security Researcher’s Paid Course Built a 2.1-Million-Credential WordPress Botnet

Posted in Commentary with tags on July 24, 2026 by itnerd

Today, SOCRadar published new research WP Botnet Master: How a Security Researcher’s Paid Course Built a 2.1-Million-Credential WordPress Botnet.

On 13 July 2026, SOCRadar Researchers recovered the complete toolkit behind a distributed WordPress brute-force operation the operator called “WP Botnet Master.” We expected to be looking at the work of a single skilled attacker. What we found was a graduation project.

The server they pulled apart did not belong to a lone hacker. It belonged to a paying student of a structured, commercial “training” program run by a WordPress security researcher who sells cybercrime as a course – complete with a curriculum, a lab blueprint, a community, and an AI-assisted workflow that lets students build and run credential-harvesting botnets with almost no skill of their own.

One student, acting alone, harvested 2,118,764 WordPress administrator credentials from 606,591 domains across 100 countries. There are roughly 295 more people in the community that trained him. The botnet is a symptom. The academy is the disease.

Key Points:

  • A threat actor operating as “KING” (@Real_King_Engine) sells a paid course, the ISAL Framework, that teaches students to stand up attack infrastructure, generate exploits with commercial AI assistants, deploy web shells, and run a credential-harvesting botnet at internet scale.
  • KING is a WPScan-credited vulnerability researcher with three published advisories and a Wordfence Intelligence researcher account carrying an approved bounty payout. These are real, verifiable identities – used as legal cover (“educational and defensive research only”) and as a credibility funnel to convert hobbyists into paying students.
  • The recovered botnet server does not belong to KING. It belongs to one of his students, a self-published developer who identifies publicly as Saeful Rochim (“dalung,” github.com/dalungid), tied to the recovered toolkit by a confirmed code-authorship fingerprint match.
  • The course teaches push-button, AI-assisted exploitation. In a paying student’s own words: “The system did everything automatically – I only drank soda.” Both Anthropic Claude and Google Gemini appear in the toolchain.
  • The output SOCRadar recovered: 272 million sites scanned, 2,118,764 administrator credentials harvested across 606,591 domains in 100 countries, and 137 active web shells across 24 countries.
  • As of the time of writing, the master command-and-control server (217.216.72.31) remained online and continued ingesting fresh target lists.

This is scalable, repeatable, and deliberately deniable cybercrime. Each of the ~295 community members is a candidate to reproduce the full operation – and an English-language edition of the course is already in development.

To view the full report, please see WP Botnet Master: How a Security Researcher’s Paid Course Built a 2.1-Million-Credential WordPress Botnet 

Anubis Claims To Have Pwned Fairlife/Coca-Cola

Posted in Commentary with tags on July 24, 2026 by itnerd

Another day. Another company pwned. Rather than go through the story I’ll let this report tell you:

Hacking gang Anubis claimed credit on Tuesday for an attack on Coca-Cola-owned dairy company fairlife, threatening to publish stolen data unless it received an unspecified ransom.

The group made the claim on its dark web site, saying it had stolen 1 terabyte of data from fairlife.

Coca-Cola did not immediately respond to a request for comment, and the hackers did not immediately return a message.

Arvind Parthasarathi, CEO and founder, CYGNVS had this comment:

“AI has fundamentally changed the scale and success of cyberattacks. Instead of manually looking for truffles in a forest, imagine an army of truffle pigs that are searching every square inch of the forest. Where attackers once had to hunt manually for vulnerabilities, today’s frontier AI models give them an army of infinitely scalable AI, constantly digging and searching for weaknesses across every corner of an organization’s environment, while also creating the exploit, taking advantage, and executing it. 

That means organizations should expect more successful attacks, more major incidents, more simultaneous incidents, and far greater operational disruption than security teams have historically planned for. 

Major cyber incidents used to be treated like once-in-a-decade hurricanes, something that only happened to someone else. Today, many CISOs are dealing with serious incidents every couple of months, and some organizations are managing multiple incidents at the same time. 

The question has shifted from ‘will you have a major incident?’ to ‘how do you treat a major incident like business as usual?’ 

Major incidents like the Coca-Cola and Fairlife attack reinforce that resilience is the key muscle. 

Organizations need an out-of-band command center that brings together security, IT, legal, risk, communications, executive leadership and trusted external partners, with pre-defined playbooks covering technical recovery, regulatory reporting, customer communications and evidence preservation. Organizations build resilience more quickly if they have already practiced these scenarios through regular tabletop exercises. Cyber resilience is about making incident response and recovery a repeatable business process.”

This should be a warning to organizations. It’s a matter of when not if you will get pwned. The question is how will you react when that happens.

The Claude Cowork flaw isn’t a patch problem, it’s a governance problem

Posted in Commentary with tags on July 23, 2026 by itnerd

Claude Cowork has a major issue that could let an AI agent act outside its intended boundaries is the latest example of a pattern the security and compliance community keeps circling back to: we’re deploying autonomous agents faster than we’re building the evidence trail to govern what they actually do.

If you are not familiar with this jailbreak, this will catch you up: Critical Security Flaw in Anthropic’s Claude Cowork Allows File Exfiltration via Prompt Injection | LinkedIn

Justin Beals, CEO & Founder of Strike Graph, an AI-native GRC and compliance automation platform had this to say:

“Every agentic AI vulnerability disclosure tells the same story. We built these tools to act on our behalf, then forgot to build the evidence trail for what they actually did. This isn’t a bug in one product. It’s a category problem.

Once an agent can take action inside a system, it’s not a feature anymore. It’s an identity. And most organizations still govern it like a checkbox on a vendor questionnaire instead of a live risk surface that needs continuous verification.

The companies that get burned by the next version of this flaw will be the ones still treating AI agent oversight as a one-time approval. Continuous evidence of what an agent did, not just what it was authorized to do, is the only model that scales.”

It would really be nice if AI was treated more strictly. But sadly it isn’t and that will come back to bite us all sooner rather than later.