Hugging Face autonomous AI agent breach

Adversaries may execute active reconnaissance scans to gather information that can be used during targeting. Active scans are those where the adversary probes victim infrastructure via network traffic, as opposed to other forms of reconnaissance that do not involve direct interaction.

Evidence: 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · ATT&CK page ↗

Adversaries may scan victims for vulnerabilities that can be used during targeting. Vulnerability scans typically check if the configuration of a target host/application (ex: software and version) potentially aligns with the target of a specific exploit the adversary may seek to use.

Evidence: 2026-09-06/openai-dsewiki-agent-collusion-egress-bypass-nondisclosure · 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · ATT&CK page ↗

Adversaries may create and cultivate social media accounts that can be used during targeting. Adversaries can create social media accounts that can be used to build a persona to further operations. Persona development consists of the development of public information, presence, history and appropriate affiliations.

Evidence: 2026-08-05/aisi-openai-cyber-range-unsanctioned-agent-actions · ATT&CK page ↗

Adversaries may obtain and abuse credentials of existing accounts as a means of gaining Initial Access, Persistence, Privilege Escalation, or Defense Evasion. Compromised credentials may be used to bypass access controls placed on various resources on systems within the network and may even be used for persistent access to remote systems and externally available services, such as VPNs, Outlook Web Access, network devices, and remote desktop. Compromised credentials may also grant an adversary increased privilege to specific systems or access to restricted areas of the network. Adversaries may choose not to use malware or tools in conjunction with the legitimate access those credentials provide to make it harder to detect their presence.

Evidence: 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Valid accounts in cloud environments may allow adversaries to perform actions to achieve Initial Access, Persistence, Privilege Escalation, or Defense Evasion. Cloud accounts are those created and configured by an organization for use by users, remote support, services, or for administration of resources within a cloud service provider or SaaS application. Cloud Accounts can exist solely in the cloud; alternatively, they may be hybrid-joined between on-premises systems and the cloud through syncing or federation with other identity sources such as Windows Active Directory.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may attempt to exploit a weakness in an Internet-facing host or system to initially access a network. The weakness in the system can be a software bug, a temporary glitch, or a misconfiguration.

Evidence: 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · 2026-05-14/gemstuffer-rubygems-weaponised-as-a-one-way-exfiltration-cha · ATT&CK page ↗

Adversaries may manipulate application software prior to receipt by a final consumer for the purpose of data or system compromise. Supply chain compromise of software can take place in a number of ways, including manipulation of the application source code, manipulation of the update/distribution mechanism for that software, or replacing compiled releases with a modified version.

Evidence: 2026-08-05/aisi-openai-cyber-range-unsanctioned-agent-actions · 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · 2026-05-14/gemstuffer-rubygems-weaponised-as-a-one-way-exfiltration-cha · ATT&CK page ↗

Adversaries may abuse command and script interpreters to execute commands, scripts, or binaries. These interfaces and languages provide ways of interacting with computer systems and are a common feature across many different platforms. Most systems come with some built-in command-line interface and scripting capabilities, for example, macOS and Linux distributions include some flavor of Unix Shell while Windows installations include the Windows Command Shell and PowerShell.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may abuse Unix shell commands and scripts for execution. Unix shells are the primary command prompt on Linux, macOS, and ESXi systems, though many variations of the Unix shell exist (e.g. sh, ash, bash, zsh, etc.) depending on the specific OS or distribution. Unix shells can control every aspect of a system, with certain commands requiring elevated privileges.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may obtain and abuse credentials of existing accounts as a means of gaining Initial Access, Persistence, Privilege Escalation, or Defense Evasion. Compromised credentials may be used to bypass access controls placed on various resources on systems within the network and may even be used for persistent access to remote systems and externally available services, such as VPNs, Outlook Web Access, network devices, and remote desktop. Compromised credentials may also grant an adversary increased privilege to specific systems or access to restricted areas of the network. Adversaries may choose not to use malware or tools in conjunction with the legitimate access those credentials provide to make it harder to detect their presence.

Evidence: 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Valid accounts in cloud environments may allow adversaries to perform actions to achieve Initial Access, Persistence, Privilege Escalation, or Defense Evasion. Cloud accounts are those created and configured by an organization for use by users, remote support, services, or for administration of resources within a cloud service provider or SaaS application. Cloud Accounts can exist solely in the cloud; alternatively, they may be hybrid-joined between on-premises systems and the cloud through syncing or federation with other identity sources such as Windows Active Directory.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may manipulate accounts to maintain and/or elevate access to victim systems. Account manipulation may consist of any action that preserves or modifies adversary access to a compromised account, such as modifying credentials or permission groups. These actions could also include account activity designed to subvert security policies, such as performing iterative password updates to bypass password duration policies and preserve the life of compromised credentials.

Evidence: 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · ATT&CK page ↗

Adversaries may add adversary-controlled credentials to a cloud account to maintain persistent access to victim accounts and instances within the environment.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may exploit software vulnerabilities in an attempt to elevate privileges. Exploitation of a software vulnerability occurs when an adversary takes advantage of a programming error in a program, service, or within the operating system software or kernel itself to execute adversary-controlled code. Security constructs such as permission levels will often hinder access to information and use of certain techniques, so adversaries will likely need to perform privilege escalation to include use of software exploitation to circumvent those restrictions.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may obtain and abuse credentials of existing accounts as a means of gaining Initial Access, Persistence, Privilege Escalation, or Defense Evasion. Compromised credentials may be used to bypass access controls placed on various resources on systems within the network and may even be used for persistent access to remote systems and externally available services, such as VPNs, Outlook Web Access, network devices, and remote desktop. Compromised credentials may also grant an adversary increased privilege to specific systems or access to restricted areas of the network. Adversaries may choose not to use malware or tools in conjunction with the legitimate access those credentials provide to make it harder to detect their presence.

Evidence: 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Valid accounts in cloud environments may allow adversaries to perform actions to achieve Initial Access, Persistence, Privilege Escalation, or Defense Evasion. Cloud accounts are those created and configured by an organization for use by users, remote support, services, or for administration of resources within a cloud service provider or SaaS application. Cloud Accounts can exist solely in the cloud; alternatively, they may be hybrid-joined between on-premises systems and the cloud through syncing or federation with other identity sources such as Windows Active Directory.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may manipulate accounts to maintain and/or elevate access to victim systems. Account manipulation may consist of any action that preserves or modifies adversary access to a compromised account, such as modifying credentials or permission groups. These actions could also include account activity designed to subvert security policies, such as performing iterative password updates to bypass password duration policies and preserve the life of compromised credentials.

Evidence: 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · ATT&CK page ↗

Adversaries may add adversary-controlled credentials to a cloud account to maintain persistent access to victim accounts and instances within the environment.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may break out of a container or virtualized environment to gain access to the underlying host. This can allow an adversary access to other containerized or virtualized resources from the host level or to the host itself. In principle, containerized / virtualized resources should provide a clear separation of application functionality and be isolated from the host environment.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may attempt to make an executable or file difficult to discover or analyze by encrypting, encoding, or otherwise obfuscating its contents on the system or in transit. This is common behavior that can be used across different platforms and the network to evade defenses.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may obtain and abuse credentials of existing accounts as a means of gaining Initial Access, Persistence, Privilege Escalation, or Defense Evasion. Compromised credentials may be used to bypass access controls placed on various resources on systems within the network and may even be used for persistent access to remote systems and externally available services, such as VPNs, Outlook Web Access, network devices, and remote desktop. Compromised credentials may also grant an adversary increased privilege to specific systems or access to restricted areas of the network. Adversaries may choose not to use malware or tools in conjunction with the legitimate access those credentials provide to make it harder to detect their presence.

Evidence: 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Valid accounts in cloud environments may allow adversaries to perform actions to achieve Initial Access, Persistence, Privilege Escalation, or Defense Evasion. Cloud accounts are those created and configured by an organization for use by users, remote support, services, or for administration of resources within a cloud service provider or SaaS application. Cloud Accounts can exist solely in the cloud; alternatively, they may be hybrid-joined between on-premises systems and the cloud through syncing or federation with other identity sources such as Windows Active Directory.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may impersonate a trusted person or organization in order to persuade and trick a target into performing some action on their behalf. For example, adversaries may communicate with victims (via Phishing for Information, Phishing, or Internal Spearphishing) while impersonating a known sender such as an executive, colleague, or third-party vendor. Established trust can then be leveraged to accomplish an adversary’s ultimate goals, possibly against multiple victims.

Evidence: 2026-09-06/openai-dsewiki-agent-collusion-egress-bypass-nondisclosure · ATT&CK page ↗

Adversaries may search compromised systems to find and obtain insecurely stored credentials. These credentials can be stored and/or misplaced in many locations on a system, including plaintext files (e.g. Shell History), operating system or application-specific repositories (e.g. Credentials in Registry), or other specialized files/artifacts (e.g. Private Keys).

Evidence: 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may search local file systems and remote file shares for files containing insecurely stored credentials. These can be files created by users to store their own credentials, shared credential stores for a group of individuals, configuration files containing passwords for a system or service, or source code/binary files containing embedded passwords.

Evidence: 2026-05-14/gemstuffer-rubygems-weaponised-as-a-one-way-exfiltration-cha · ATT&CK page ↗

Adversaries may attempt to access the Cloud Instance Metadata API to collect credentials and other sensitive data.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may attempt to discover containers and other resources that are available within a containers environment. Other resources may include images, deployments, pods, nodes, and other information such as the status of a cluster.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may exploit remote services to gain unauthorized access to internal systems once inside of a network. Exploitation of a software vulnerability occurs when an adversary takes advantage of a programming error in a program, service, or within the operating system software or kernel itself to execute adversary-controlled code. A common goal for post-compromise exploitation of remote services is for lateral movement to enable access to a remote system.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may search local system sources, such as file systems, configuration files, local databases, virtual machine files, or process memory, to find files of interest and sensitive data prior to Exfiltration.

Evidence: 2026-07-31/anthropic-cyber-eval-environment-escape-pypi-package · ATT&CK page ↗

Adversaries may communicate using OSI application layer protocols to avoid detection/network filtering by blending in with existing traffic. Commands to the remote system, and often the results of those commands, will be embedded within the protocol traffic between the client and server.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may use a connection proxy to direct network traffic between systems or act as an intermediary for network communications to a command and control server to avoid direct connections to their infrastructure. Many tools exist that enable traffic redirection through proxies or port redirection, including HTRAN, ZXProxy, and ZXPortMap. Adversaries use these types of proxies to manage command and control communications, reduce the number of simultaneous outbound network connections, provide resiliency in the face of connection loss, or to ride over existing trusted communications paths between victims to avoid suspicion. Adversaries may chain together multiple proxies to further disguise the source of malicious traffic.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may use an external proxy to act as an intermediary for network communications to a command and control server to avoid direct connections to their infrastructure. Many tools exist that enable traffic redirection through proxies or port redirection, including HTRAN, ZXProxy, and ZXPortMap. Adversaries use these types of proxies to manage command and control communications, to provide resiliency in the face of connection loss, or to ride over existing trusted communications paths to avoid suspicion.

Evidence: 2026-09-06/openai-dsewiki-agent-collusion-egress-bypass-nondisclosure · ATT&CK page ↗

Adversaries may take advantage of routing schemes in Content Delivery Networks (CDNs) and other services which host multiple domains to obfuscate the intended destination of HTTPS traffic or traffic tunneled through HTTPS. Domain fronting involves using different domain names in the SNI field of the TLS header and the Host field of the HTTP header. If both domains are served from the same CDN, then the CDN may route to the address specified in the HTTP header after unwrapping the TLS header. A variation of the the technique, "domainless" fronting, utilizes a SNI field that is left blank; this may allow the fronting to work even when the CDN attempts to validate that the SNI and HTTP Host fields match (if the blank SNI fields are ignored).

Evidence: 2026-09-06/openai-dsewiki-agent-collusion-egress-bypass-nondisclosure · ATT&CK page ↗

Adversaries may use an existing, legitimate external Web service as a means for relaying data to/from a compromised system. Popular websites, cloud services, and social media acting as a mechanism for C2 may give a significant amount of cover due to the likelihood that hosts within a network are already communicating with them prior to a compromise. Using common services, such as those offered by Google, Microsoft, or Twitter, makes it easier for adversaries to hide in expected noise. Web service providers commonly use SSL/TLS encryption, giving adversaries an added level of protection.

Evidence: 2026-07-21/hugging-face-autonomous-ai-agent-production-breach · ATT&CK page ↗

Adversaries may use an existing, legitimate external Web service as a means for sending commands to and receiving output from a compromised system over the Web service channel. Compromised systems may leverage popular websites and social media to host command and control (C2) instructions. Those infected systems can then send the output from those commands back over that Web service channel. The return traffic may occur in a variety of ways, depending on the Web service being utilized. For example, the return traffic may take the form of the compromised system posting a comment on a forum, issuing a pull request to development project, updating a document hosted on a Web service, or by sending a Tweet.

Evidence: 2026-09-06/openai-dsewiki-agent-collusion-egress-bypass-nondisclosure · ATT&CK page ↗

Adversaries may tunnel network communications to and from a victim system within a separate protocol to avoid detection/network filtering and/or enable access to otherwise unreachable systems. Tunneling involves explicitly encapsulating a protocol within another. This behavior may conceal malicious traffic by blending in with existing traffic and/or provide an outer layer of encryption (similar to a VPN). Tunneling could also enable routing of network packets that would otherwise not reach their intended destination, such as SMB, RDP, or other traffic that would be filtered by network appliances or not routed over the Internet.

Evidence: 2026-08-05/aisi-openai-cyber-range-unsanctioned-agent-actions · ATT&CK page ↗

Adversaries may exfiltrate data to a webhook endpoint rather than over their primary command and control channel. Webhooks are simple mechanisms for allowing a server to push data over HTTP/S to a client without the need for the client to continuously poll the server. Many public and commercial services, such as Discord, Slack, and `webhook.site`, support the creation of webhook endpoints that can be used by other services, such as Github, Jira, or Trello. When changes happen in the linked services (such as pushing a repository update or modifying a ticket), these services will automatically post the data to the webhook endpoint for use by the consuming application.

Evidence: 2026-05-14/gemstuffer-rubygems-weaponised-as-a-one-way-exfiltration-cha · ATT&CK page ↗