If you want AI to tell you a story, it will. If you want that story to sound like one of your favorite authors, it can. And if you’re one of the authors that AI can imitate, you might be a little upset at what feels like theft.
In 2024, the authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson sued Anthropic, the creator of Claude, alleging that the company had wrongfully digested millions of copyrighted works—including some of their very own—to train its AI models. The lawsuit grew to include more than 300,000 writers, and in September 2025, Anthropic agreed to pay $1.5 billion to settle the claims.
That headline-worthy payout, however, would eventually be paired with more startling news.
In January 2026, a district court judge unsealed thousands of documents related to the litigation. When The Washington Post investigated the documents, reporters found references to a secret project inside Anthropic called “Project Panama,” which Anthropic itself described as the company’s effort “to destructively scan all the books in the world.”
Anthropic is not alone in this.
On August 17, 404 Media co-founder and reporter Emanuel Maiberg revealed that Amazon is doing something similar inside a department it calls VGT3. By hiding an Apple AirTag in one book from a 1,000-book order, and then tracking the shipment across the country, 404 Media identified the book’s final destination to be a facility outside Las Vegas, where, according to employees, books are cut apart and scanned.
Today, on the Lock and Code podcast with host David Ruiz, we speak with Maiberg about Amazon’s VGT3 operation and the likely commonplace practice of AI companies purchasing, scanning, and destroying books in an effort to build “frontier” models.
“They’re buying a book, they’re scanning it, they’re mulching the book, and then the digital version of the book exists behind this wall where we don’t even know if they plan to sell it, right? It’s like they might keep it behind a wall, and the only way we see it come out again from behind that wall is in the form of an answer from a chatbot.”
The news coverage surrounding the breach of the AI platform Hugging Face by OpenAI’s autonomous agents reads like a scene straight out of a Terminator movie (apparently, a prequel). Yet for enterprise security teams — even those that don’t develop or deploy artificial intelligence — the detailed incident report published by Hugging Face provides a critical operational roadmap. The incident needs to be evaluated on the basis of a single, fundamental question: “Which organizational security flaws, missing controls, configurations, or technologies allowed this attack to succeed?”
What happened?
During cyber-capability testing of OpenAI’s advanced AI models against the ExploitGym benchmark, an agent found a way to break out of its confinement, reached the internet, and breached the infrastructure of several companies — one of them being Hugging Face. The model reasoned that it could find solutions to the benchmark challenges on these external systems. Dozens of articles have since covered these incidents in detail, so no need to go over that here. What truly matters is what transpired within Hugging Face’s systems between July 9 and 13 while the AI agent operated unchecked inside. The Hugging Face incident report includes an interactive attack timeline, so we’ll just list the key stages. A malicious configuration in a dataset uploaded by the agent allowed it to exfiltrate credentials for a worker pod, and execute code within it. The agent then exploited cloud environment metadata and escaped the privileged pod to obtain root-level access on the host. It retrieved a substantial number of secrets from storage and used a stolen VPN key and shared cluster administrator credentials to pivot deeper into the internal network and source code repository before the Hugging Face security team blocked its access.
How significant is the fact that this was next-generation AI?
Regardless of the specific area of activity of a company, tomorrow it may become a target of human attackers, fully autonomous systems, or AI agents manually controlled by human hackers.
Many AI practitioners have rushed to dismiss this incident based on assumptions that OpenAI will never allow its tech to get into serious hacking. However, security researchers have argued for months that even currently available open-source models can be used to launch both offensive research and real-world attacks. Human oversight, while still a requirement, can be minimal. Even mid-tier models can easily iterate through a dozen vulnerabilities, inventory a compromised infrastructure, and quickly figure out how to expand their reach across a network. Defenders must analyze incidents like this in detail, map the key findings to their own environments, and adapt security controls to counter both stealthy human actors and rapid, noisy AI agents.
Three distinguishing features of AI-powered attacks
The techniques detailed in the Hugging Face report are not novel — the agent didn’t invent anything new. However, three factors regarding the attack’s speed and scale fundamentally shift the economics for both attackers and defenders:
Trial and error becomes significantly cheaper. Where a human hacker might test five attack vectors, an agent can iterate through five hundred. Scenarios that security teams previously dismissed as “theoretically possible, but impractical and unlikely” and placed them on the backlog can now become active threats.
Attack execution and incident response windows shrink. High-speed attacks are not unheard of, while some ransomware operators achieve compromise within hours even without AI. However, this accelerated pace will likely become the new baseline in cybersecurity, as demonstrated by the agent obtaining admin privileges at Hugging Face within 13 hours.
Numerous alerts and log entries generated by the agent’s trial-and-error attempts can both help and hinder attackers as well as defenders. This noise can serve both as an indicator for threat detection, and as camouflage to hide meaningful actions. It also acts as a natural bottleneck for incident response when the workflows are fully manual.
How to protect corporate infrastructure
Don’t ignore alerts from triggered defensive controls. This classic mistake — often highlighted by our incident response experts — was made at Hugging Face as well. Malware execution, port scanning, WAF triggers, requests to non-allowlisted resources, and failed authentication attempts are likely some of the event types that are blocked automatically. However, all these events are often categorized by SIEM systems as purely informational and routinely ignored. In the Hugging Face incident, numerous server-side request forgery (SSRF) attack attempts were successfully blocked over several days, yet no one batted an eyelash. Much like persistent human adversaries, the AI agent continuously retried until it succeeded. To detect such attacks, establish clear rules and triage workflows for defensive alert telemetry to ensure that priority automatically escalates when event volume spikes.
Investigate rapidly escalating error counts. Legitimate operations rarely generate hundreds of authentication failures within 10 minutes across four distinct systems. Analyze historical baselines, and generate medium or high-severity alerts depending on the deviation from normal error rates. As part of automated response protocols, you can temporarily block the user or system that is generating the anomalous activity, or enforce rate-limiting on them.
Configure escalation paths and conduct security team drills. Hugging Face’s correlation engine successfully detected the attack, yet it failed to grab the human operators’ attention. During security drills, organization must evaluate how quickly key personnel respond to critical incidents — even when these happen during off-hours such as at night on weekends.
Isolate services — internal ones included — that process untrusted input by design. The primary vulnerability at the start of the attack was a node configured to process machine learning datasets, which are widely known for potentially containing executable code. This node, capable of executing untrusted code, operated without a sandbox or adequate isolation controls. The risk extends beyond machine learning — to document conversion engines, image processing pipelines, user file upload handlers, and any other server-side processes that perform deserialization or template rendering. Any server-side execution of this nature must run under maximum feasible isolation, restricted from accessing external or internal network hosts beyond an absolute minimum, and ideally enclosed within a sandbox.
Restrictegress. Most organizational servers don’t require unrestricted connectivity to arbitrary external hosts. If implementing strict default-deny egress policies presents technical challenges, consider a streamlined approach: enforce DNS traffic resolution exclusively through your enterprise DNS server, and configure it to block known public services that are commonly abused as command-and-control infrastructure or exfiltration channels such as Pastebin or OneDrive. Similar outbound restrictions can also be enforced via an NGFW.
Avoid issuing large batches of secrets, or storing them as a single object. Even with secret managers in place, an attacker can gain broad access if a compromised account holds excessive permissions. In the Hugging Face breach, a single Kubernetes Secret object contained 136 keys. The fewer secrets issued, the lower the impact of a compromise — and the easier key-rotation becomes.
Avoid long-lived secrets and don’t store them in environment variables that are inherited by all child processes.
Isolate applications from metadata servers. Most cloud environments run services — such as the EC2 Instance Metadata Service — that issue credentials and system details upon request. Isolating these from arbitrary applications — via firewall rules or by enforcing IMDSv2 — eliminates the most direct path from a single application compromise to a full account takeover.
Audit server and application identities. Service accounts with broad access, OAuth tokens, cross-environment keys, and other artifacts granting excessive privileges are most frequently found in server and app identities rather than human user accounts.
Track usage of credentials at anomalous entry points. In the Hugging Face attack, credentials assigned to an internal host were leveraged from an external one. Detection rules that flag these anomalies are straightforward to implement and deliver high-fidelity, low-noise signals of an attack that’s already progressed beyond its initial stages.
Use short-lived, single-use keys when provisioning new devices and services. During the incident, a single key was used to register 181 separate devices without triggering a single alert.
Conventional security awareness initiatives often fall short because they fail to instill lasting behavioral change. Access our comprehensive research article for an in-depth exploration of the underlying causes and actionable remediation strategies.
Things got a bit hectic and I didn’t get the January training news monthly blog post out so I’m combining two months in this post. The following contains newly released Digital Forensics, Incident Response,...
This blog post is aimed at the intermediate level learner in the fields of data science and artificial intelligence. If you would like to read up on some fundamentals, here […]
The following contains newly released Digital Forensics, Incident Response, Malware Analysis and OSINT training and tools from November, as well as upcoming live online training and events for December. Like the Free & Affordable Training...
The following contains newly released Digital Forensics, Incident Response, Malware Analysis and OSINT training and tools from October, as well as upcoming live online training and events for November. Like the Free & Affordable Training...
The following contains newly released Digital Forensics, Incident Response, Malware Analysis and OSINT training and tools from September, as well as upcoming live online training and events for October. Like the Free & Affordable Training...
| Carrie Roberts // Guest Author Carrie Roberts is an Antisyphon instructor and experienced cyber security professional who has mentored many on their journey into cyber. My name is Carrie […]
In this BHIS podcast, originally recorded as a live webcast, we cover some new techniques and tactics on how to track attackers via various honey tokens. We cover how to […]
Chevy Swanson // I got my start in InfoSec through a few competitions during my time in high school. My team and I were fortunate to have a supportive school and […]
Login
