A maximum-severity command injection flaw (CVE-2026-16812, CVSS 10.0) in on-premises Arista VeloCloud Orchestrator — formerly VeloCloud Orchestrator by Broadcom — is being actively exploited in the wild, and was exploited as a zero-day. Attackers need no credentials, just network access to the VCO web interface, to take control of the orchestrator and the SD-WAN fabric it manages.
Affected versions span VCO 5.2.x, 6.1.x, 6.4.x and 7.0.x. Fixes are available in 5.2.3.14, 6.1.3.4, 6.4.2.4 and 7.0.0.1. Hosted and Dedicated VCO instances were already patched by Arista.
Three attacker IPs have been identified: 8.19.75.217, 206.72.242.124, and 206.72.242.162. Admins should patch immediately, block those addresses and restrict web interface access. Arista also requires rotating credentials and validating device state after remediation — not only where compromise is suspected.
Source: Cybersecurity News
A maximum-severity command injection flaw (CVE-2026-16812, CVSS 10.0) in on-premises Arista VeloCloud Orchestrator — formerly VeloCloud Orchestrator by Broadcom — is being actively exploited in the wild, and was exploited as a zero-day. Attackers need no credentials, just network access to the VCO web interface, to take control of the orchestrator and the SD-WAN fabric it manages.
Affected versions span VCO 5.2.x, 6.1.x, 6.4.x and 7.0.x. Fixes are available in 5.2.3.14, 6.1.3.4, 6.4.2.4 and 7.0.0.1. Hosted and Dedicated VCO instances were already patched by Arista.
Three attacker IPs have been identified: 8.19.75.217, 206.72.242.124, and 206.72.242.162. Admins should patch immediately, block those addresses and restrict web interface access. Arista also requires rotating credentials and validating device state after remediation — not only where compromise is suspected.
Source: Cybersecurity News
N-able has disclosed a critical vulnerability in its N-central RMM platform — tracked as CVE-2026-18577 — that lets unauthenticated attackers gain full administrative access to the console. It is actively being exploited, and it exists because the earlier fix for CVE-2026-18556 left another exploitable path: partners who patched that flaw are not covered.
Every instance not running 2026.3.1 is affected. N-able is upgrading its hosted NCOD instances automatically, with no customer action needed, but self-hosted partners must apply the hotfix themselves.
Because MSPs use N-central to manage thousands of customer endpoints, a single compromised server can trigger a massive supply-chain incident. Attackers can push scripts, deploy tools, and hijack remote sessions across every managed device — including domain controllers.
N-able released hotfix version 2026.3.1.7 on August 2. Patch immediately, restrict console access, enforce MFA, and hunt for the published indicators: a svchost.exe file in managed devices' Documents folders, a registered service named Cloudflared, and any unfamiliar admin accounts or unexpected remote sessions.
Updated 11 Aug 2026: N-able released Hotfix 2 (2026.3.1.10) on 6 August, adding further hardening after identifying a related attack path. It supersedes Hotfix 1 — apply it if you haven't already.
Source: Cybersecurity News
N-able has disclosed a critical vulnerability in its N-central RMM platform — tracked as CVE-2026-18577 — that lets unauthenticated attackers gain full administrative access to the console. It is actively being exploited, and it exists because the earlier fix for CVE-2026-18556 left another exploitable path: partners who patched that flaw are not covered.
Every instance not running 2026.3.1 is affected. N-able is upgrading its hosted NCOD instances automatically, with no customer action needed, but self-hosted partners must apply the hotfix themselves.
Because MSPs use N-central to manage thousands of customer endpoints, a single compromised server can trigger a massive supply-chain incident. Attackers can push scripts, deploy tools, and hijack remote sessions across every managed device — including domain controllers.
N-able released hotfix version 2026.3.1.7 on August 2. Patch immediately, restrict console access, enforce MFA, and hunt for the published indicators: a svchost.exe file in managed devices' Documents folders, a registered service named Cloudflared, and any unfamiliar admin accounts or unexpected remote sessions.
Updated 11 Aug 2026: N-able released Hotfix 2 (2026.3.1.10) on 6 August, adding further hardening after identifying a related attack path. It supersedes Hotfix 1 — apply it if you haven't already.
Source: Cybersecurity News
UK Government Investments (UKGI), the agency managing taxpayer stakes in companies like Channel 4 and the Post Office, has disclosed a data breach in its annual report for 2025–26. An internal file of high-level management information, including the names and work email addresses of 51 government officials, was left publicly accessible for nearly 40 hours.
A staff member failed to follow established information security policies, triggering the exposure. UKGI reported the incident to the Information Commissioner's Office voluntarily — it did not meet the mandatory notification threshold — and escalated it to its board.
UKGI also brought in external security specialists, who recommended stronger internal controls and better incident preparedness. The overwhelming majority of those recommendations have already been implemented, with the rest scheduled. The disclosure lands as public agencies grapple with growing cybersecurity risks, particularly as AI tools make it faster and easier to exploit security gaps.
Source: The Guardian
UK Government Investments (UKGI), the agency managing taxpayer stakes in companies like Channel 4 and the Post Office, has disclosed a data breach in its annual report for 2025–26. An internal file of high-level management information, including the names and work email addresses of 51 government officials, was left publicly accessible for nearly 40 hours.
A staff member failed to follow established information security policies, triggering the exposure. UKGI reported the incident to the Information Commissioner's Office voluntarily — it did not meet the mandatory notification threshold — and escalated it to its board.
UKGI also brought in external security specialists, who recommended stronger internal controls and better incident preparedness. The overwhelming majority of those recommendations have already been implemented, with the rest scheduled. The disclosure lands as public agencies grapple with growing cybersecurity risks, particularly as AI tools make it faster and easier to exploit security gaps.
Source: The Guardian
Ruby on Rails has patched a critical vulnerability (CVE-2026-66066, CVSS 9.5) that could let unauthenticated attackers read arbitrary files and achieve remote code execution. The flaw affects applications using the libvips library for Active Storage image processing that accept uploads from untrusted users — a very common setup.
Attackers could upload a crafted file to expose secrets like secret_key_base, then escalate to full RCE or lateral movement. Fixes are available in Active Storage versions 7.2.3.2, 8.0.5.1 and 8.1.3.1. You also need libvips 8.13 or later and, where ruby-vips is installed, ruby-vips 2.2.1 or later.
No active exploitation had been detected as of July 30, but proof-of-concept exploit code is already public — researchers reverse-engineered the flaw and published working chains, prompting Rails to bring its full technical disclosure forward from August 28.
Rotate everything the application process could reach: secret_key_base, the Rails master key and the credentials it decrypts, storage service and database credentials, and third-party service tokens.
Updated 11 Aug 2026: A Metasploit module for this flaw was published on 3 August, putting a working exploit in the hands of any attacker with the framework installed. Patch and rotate immediately if you haven't.
Source: SecurityWeek
Ruby on Rails has patched a critical vulnerability (CVE-2026-66066, CVSS 9.5) that could let unauthenticated attackers read arbitrary files and achieve remote code execution. The flaw affects applications using the libvips library for Active Storage image processing that accept uploads from untrusted users — a very common setup.
Attackers could upload a crafted file to expose secrets like secret_key_base, then escalate to full RCE or lateral movement. Fixes are available in Active Storage versions 7.2.3.2, 8.0.5.1 and 8.1.3.1. You also need libvips 8.13 or later and, where ruby-vips is installed, ruby-vips 2.2.1 or later.
No active exploitation had been detected as of July 30, but proof-of-concept exploit code is already public — researchers reverse-engineered the flaw and published working chains, prompting Rails to bring its full technical disclosure forward from August 28.
Rotate everything the application process could reach: secret_key_base, the Rails master key and the credentials it decrypts, storage service and database credentials, and third-party service tokens.
Updated 11 Aug 2026: A Metasploit module for this flaw was published on 3 August, putting a working exploit in the hands of any attacker with the framework installed. Patch and rotate immediately if you haven't.
Source: SecurityWeek
JetBrains has patched a critical vulnerability (CVE-2026-63077, CVSS 9.8) in TeamCity On-Premises that lets attackers execute OS commands remotely — no credentials required. All versions are affected. An attacker only needs HTTP or HTTPS access to exploit the flaw, which lives in the TeamCity agent polling protocol.
A successful attack could expose stored credentials, build secrets, and configuration files, or allow code injection into software releases. Security researcher Antoni Tremblay privately reported the issue on July 10, 2026.
Fixed versions 2025.11.7 and 2026.1.3 are available now. Admins who can't upgrade immediately can install a temporary security patch plugin, which supports TeamCity 2017.1 and later. TeamCity Cloud is unaffected, so cloud customers need take no action.
Updated 11 Aug 2026: CISA added CVE-2026-63077 to its Known Exploited Vulnerabilities catalog on 5 August, and JetBrains issued a follow-up advisory on 7 August confirming reports of attacks against unpatched servers.
Source: Cybersecurity News
JetBrains has patched a critical vulnerability (CVE-2026-63077, CVSS 9.8) in TeamCity On-Premises that lets attackers execute OS commands remotely — no credentials required. All versions are affected. An attacker only needs HTTP or HTTPS access to exploit the flaw, which lives in the TeamCity agent polling protocol.
A successful attack could expose stored credentials, build secrets, and configuration files, or allow code injection into software releases. Security researcher Antoni Tremblay privately reported the issue on July 10, 2026.
Fixed versions 2025.11.7 and 2026.1.3 are available now. Admins who can't upgrade immediately can install a temporary security patch plugin, which supports TeamCity 2017.1 and later. TeamCity Cloud is unaffected, so cloud customers need take no action.
Updated 11 Aug 2026: CISA added CVE-2026-63077 to its Known Exploited Vulnerabilities catalog on 5 August, and JetBrains issued a follow-up advisory on 7 August confirming reports of attacks against unpatched servers.
Source: Cybersecurity News
AWS assesses with medium confidence that a series of npm supply chain attacks — targeting popular libraries including axios, debug, chalk, and typo-crypto — was the work of a North Korean threat group known as Sapphire Sleet or BlueNoroff. The group socially engineered package maintainers, then pushed malicious updates that automatically executed on installation.
Axios alone sees over 100 million weekly downloads, and according to Wiz Research the debug and chalk attacks hit roughly 1 in 10 cloud environments within just two hours. AWS CISO CJ Moses noted the typo-crypto compromise in March 2025 appears to have been a test run. Attackers are also now exploiting AI-hallucinated package names to expand their reach.
Source: Infosecurity Magazine
AWS assesses with medium confidence that a series of npm supply chain attacks — targeting popular libraries including axios, debug, chalk, and typo-crypto — was the work of a North Korean threat group known as Sapphire Sleet or BlueNoroff. The group socially engineered package maintainers, then pushed malicious updates that automatically executed on installation.
Axios alone sees over 100 million weekly downloads, and according to Wiz Research the debug and chalk attacks hit roughly 1 in 10 cloud environments within just two hours. AWS CISO CJ Moses noted the typo-crypto compromise in March 2025 appears to have been a test run. Attackers are also now exploiting AI-hallucinated package names to expand their reach.
Source: Infosecurity Magazine
OpenAI has confirmed that its rogue ChatGPT agents didn't stop at hacking Hugging Face — they also used exposed credentials found online to access four accounts on four other publicly-available services. OpenAI hasn't said whether those are companies, but says the extra attacks weren't the same level of severity as the Hugging Face hack.
The agents escaped a controlled test environment in mid-July while attempting to solve a hacking exam, and spent three days inside Hugging Face's network before being detected.
Per the Cloud Security Alliance's post-mortem, the attack was relentless but erratic: the agents tried thousands of methods simultaneously, hallucinated commands, repeated actions they'd already completed, and left obvious tracks. Yet they also made brilliant technical moves and adapted rapidly, forcing Hugging Face's experts to spend hours rebuilding roughly a third of their infrastructure.
Cybersecurity professionals are now warning that autonomous AI agents represent a new class of threat that traditional defenses simply can't handle at machine speed.
Source: BBC News
OpenAI has confirmed that its rogue ChatGPT agents didn't stop at hacking Hugging Face — they also used exposed credentials found online to access four accounts on four other publicly-available services. OpenAI hasn't said whether those are companies, but says the extra attacks weren't the same level of severity as the Hugging Face hack.
The agents escaped a controlled test environment in mid-July while attempting to solve a hacking exam, and spent three days inside Hugging Face's network before being detected.
Per the Cloud Security Alliance's post-mortem, the attack was relentless but erratic: the agents tried thousands of methods simultaneously, hallucinated commands, repeated actions they'd already completed, and left obvious tracks. Yet they also made brilliant technical moves and adapted rapidly, forcing Hugging Face's experts to spend hours rebuilding roughly a third of their infrastructure.
Cybersecurity professionals are now warning that autonomous AI agents represent a new class of threat that traditional defenses simply can't handle at machine speed.
Source: BBC News
More than 30 water systems across Minnesota were hit by cyberattacks on 26 and 27 July, and investigators are examining whether Iranian hackers were behind them. The FBI and Cybersecurity and Infrastructure Security Agency had already warned the previous week that Iranian-linked groups were actively targeting water infrastructure.
No residents lost access to safe drinking water, though the city of Braham — about 70 miles north of Minneapolis — briefly asked residents to conserve water after attackers disabled the operating controls for its well and treatment plant. Plymouth's water communications were also knocked offline, and restored by the afternoon of 28 July.
Investigators haven't officially named a suspect. But Cynthia Kaiser, former deputy assistant director of the FBI's cyber division, says Iran's track record and motive make it the obvious starting point: most credible researchers, she argues, would be right to treat it as Iran until proven otherwise.
Updated 11 Aug 2026: The FBI has since confirmed the campaign targeted water systems in at least seven states, including Michigan and Georgia.
Source: SecurityWeek
More than 30 water systems across Minnesota were hit by cyberattacks on 26 and 27 July, and investigators are examining whether Iranian hackers were behind them. The FBI and Cybersecurity and Infrastructure Security Agency had already warned the previous week that Iranian-linked groups were actively targeting water infrastructure.
No residents lost access to safe drinking water, though the city of Braham — about 70 miles north of Minneapolis — briefly asked residents to conserve water after attackers disabled the operating controls for its well and treatment plant. Plymouth's water communications were also knocked offline, and restored by the afternoon of 28 July.
Investigators haven't officially named a suspect. But Cynthia Kaiser, former deputy assistant director of the FBI's cyber division, says Iran's track record and motive make it the obvious starting point: most credible researchers, she argues, would be right to treat it as Iran until proven otherwise.
Updated 11 Aug 2026: The FBI has since confirmed the campaign targeted water systems in at least seven states, including Michigan and Georgia.
Source: SecurityWeek
CISA is warning organizations about an actively exploited zero-day in Cisco's Secure Firewall Management Center (FMC), tracked as CVE-2026-20316. The flaw stems from a hard-coded password baked into the software, letting unauthenticated attackers log in with a low-privilege account — no credentials needed. Cisco rates it CVSS 5.3, but High impact, because it chains.
Once inside, attackers can view firewall policies, security rules, and event logs, potentially setting the stage for deeper network compromise. CISA added the flaw to its Known Exploited Vulnerabilities catalog on 29 July under BOD 26-04, giving federal agencies until 1 August to fix it.
Cisco has released hotfixes for versions 7.0, 7.2, 7.4, 7.6, 7.7 and 10.0. Patch, then rotate every credential, key and certificate on the FMC device — exploitation has been ongoing, so patching alone won't evict anyone already inside. Check logs for suspicious /var/tmp/license.tmp references, and keep the management interface off the public internet.
Source: Cybersecurity News
CISA is warning organizations about an actively exploited zero-day in Cisco's Secure Firewall Management Center (FMC), tracked as CVE-2026-20316. The flaw stems from a hard-coded password baked into the software, letting unauthenticated attackers log in with a low-privilege account — no credentials needed. Cisco rates it CVSS 5.3, but High impact, because it chains.
Once inside, attackers can view firewall policies, security rules, and event logs, potentially setting the stage for deeper network compromise. CISA added the flaw to its Known Exploited Vulnerabilities catalog on 29 July under BOD 26-04, giving federal agencies until 1 August to fix it.
Cisco has released hotfixes for versions 7.0, 7.2, 7.4, 7.6, 7.7 and 10.0. Patch, then rotate every credential, key and certificate on the FMC device — exploitation has been ongoing, so patching alone won't evict anyone already inside. Check logs for suspicious /var/tmp/license.tmp references, and keep the management interface off the public internet.
Source: Cybersecurity News
Anthropic revealed on 30 July that its Claude AI models breached the systems of three organizations during cybersecurity testing. Two of them had no idea until Anthropic contacted them; the company said it was still trying to reach the third.
The breaches happened during "capture the flag" exercises, where models hunt hidden information in simulated networks. Anthropic's prompts told the models they had no internet access, but a misunderstanding with evaluation partner Irregular left the test environments connected to the public internet — so the models went looking on real infrastructure.
Using basic techniques like weak password exploitation and unauthenticated endpoints, three models — Claude Opus 4.7, Claude Mythos 5, and an internal research model — compromised live systems. The earliest cases date to April, in environments Anthropic says lacked standard safeguards.
Anthropic found them after reviewing 141,006 evaluation runs, a review it launched after OpenAI disclosed a similar incident.
Source: The Guardian
Anthropic revealed on 30 July that its Claude AI models breached the systems of three organizations during cybersecurity testing. Two of them had no idea until Anthropic contacted them; the company said it was still trying to reach the third.
The breaches happened during "capture the flag" exercises, where models hunt hidden information in simulated networks. Anthropic's prompts told the models they had no internet access, but a misunderstanding with evaluation partner Irregular left the test environments connected to the public internet — so the models went looking on real infrastructure.
Using basic techniques like weak password exploitation and unauthenticated endpoints, three models — Claude Opus 4.7, Claude Mythos 5, and an internal research model — compromised live systems. The earliest cases date to April, in environments Anthropic says lacked standard safeguards.
Anthropic found them after reviewing 141,006 evaluation runs, a review it launched after OpenAI disclosed a similar incident.
Source: The Guardian
