IT eBulletins
Search and filter sharp, essential coverage across every IT topic that matters.
Coding agents shared screenshots through public repositories outside corporate monitoring, exposing billing records and unreleased features.
A PAN-OS management interface flaw grants root on the firewall, has no workaround, and came with a three-day federal deadline.
AI-assisted discovery has pushed Microsoft's monthly fixes from 137 in May to a record 1,411, while the teams deploying them still work to a 55-day median.
A subscription Mac stealer harvests the cloud and code credentials on developer laptops, then stays behind as a remote shell that relaunches every hour.
Only 2% of organizations govern their AI agents with purpose-built identity tools, while the gateways and MCP servers beneath them leak keys and accept forged tokens.
Qilin and Settra are opening VPN sessions through a GlobalProtect authentication bypass, and a patched gateway can leave those sessions open.
Push Security found ClickFix behind 67% of its detections in August, with 4 in 5 payloads reached from a search engine.
PwC found 84% of leaders expect bigger cyber budgets, yet half say attacks on AI systems are the threat they are least ready for.
Comparitech counted a record 2,627 ransomware attacks in Q3 2026, but only 247 were confirmed by the organizations named.
Gigamon found 80% of CISOs name unsanctioned AI their top data challenge, and Gartner says 54% have no defined way to limit what AI agents can reach.
Microsoft puts median time to weaponization below 24 hours, and Sysdig documented JADEPUFFER, the first ransomware operation run end to end by an AI agent.
ISACA found only 8% of organizations regularly rehearse an AI incident, and 48% lack an AI playbook or do not know if one exists. DigiCert and Kiteworks show governance trailing deployment in the same way.
Zscaler's ThreatLabz counted 896.2 terabytes stolen by the ten largest ransomware groups, up 275.8% in a year, with 62% of named victims at manager level or above. Arctic Wolf and Coveware both report crews skipping encryption to sell silence instead.
Cisco surveyed 8,000 security professionals and found only 8% of organizations in its top tier and just 21% able to deploy a new security control within six months of approval. Internal friction carried the most weight in its scoring.
SpyCloud's survey of 750 security leaders found organizations without visibility into stolen session cookies had a 50% identity event rate, against 37% for those that monitor them, and 53% see malware only on managed devices. Moonlock counted 67% growth in new macOS backdoor variants as Mac stealers began staying behind.
VulnCheck found 42.5% of the edge device vulnerabilities exploited in 2025 hit end-of-life or likely end-of-life hardware, and 65% of botnet-exploited flaws targeted unsupported devices. In September, CISA gave federal agencies two days, then three, to patch exploited Cisco, Citrix, and Fortinet edge flaws.
Unit 42 found only 12 of 405 AI-enabled malware samples on a real production endpoint, with roughly 97% never leaving sandboxes and VirusTotal. Koi Security found 341 malicious AI agent skills, 335 of them delivering the Atomic macOS Stealer, as AI tools became the delivery route.
Black Kite found 76 of the 140 vendors most concentrated in financial services carry at least one CISA Known Exploited Vulnerability, and confirmed breaches among them rose from 6 to 39 in a year. Finance ransomware rose 76% in Q1 2026, and one managed service provider compromise reached 32 institutions.
Digital Realty's survey of 2,131 IT decision-makers found 40% now name a lack of specialized infrastructure as their primary AI constraint, up from 9% in 2024. Seagate found 83% call themselves prepared for AI's data demands, but only 38% say they are fully prepared for the long term.
WatchGuard found endpoint ransomware detections down 68% year over year, even as public extortion claims neared 5,000, a record. Check Point found a 51-point gap between AI security intent and enforcement, and Darktrace found 46% of teams still feel unprepared.
Cisco Talos found authentication abuse in 65% of Q2 2026 engagements, nearly double the prior quarter, while phishing crossed the halfway mark. A QR-code PDF campaign and a new ransomware variant show how far past the login screen the problem now reaches.
Cyera's review of 7,246 AI incidents found 188 cases where an autonomous agent caused real damage with no attacker involved. Google's Mandiant traced a separate agent that ran up a $50,000 cloud bill in under an hour.
CrowdStrike's 2026 Global Threat Report puts the average eCrime breakout time at 29 minutes, a 65% jump in speed. Mandiant's M-Trends 2026 puts median detection dwell time at 14 days, up from 11.
Apple's September update fixed 273 unique vulnerabilities in one release, but a macOS flaw it had already fixed in August was still being used to plant cryptocurrency miners.
Moonlock Lab found unique malicious macOS samples up roughly 40% year over year, with Odyssey and AMOS now behind more than nine in ten stealer detections on Mac.
Cyberhaven found 39.7% of AI interactions now involve sensitive data, and UpGuard found 81% of employees use AI tools their employer never approved. Underneath both sits a data estate that is already 80% unstructured and growing 55 to 65% a year.
Cisco Talos found phishing reclaimed the top spot for initial access in Q1 2026, and CrowdStrike found 82% of the intrusions that follow are now malware-free. Verizon ties it together: 73% of ransomware victims had a credential leak in the year before the attack.
A Cloud Security Alliance and Thales study of 210 IT and security professionals found most unstructured data goes unprotected, only 9% can scan it in real time, and 47% now name AI the top future threat to it.
Unit 42 found encryption present in just 78% of 2025 extortion cases as data theft alone becomes viable. The fastest attacks now exfiltrate in 72 minutes, and backup recovery is falling as a result.
Gravitee found only 21.9% of enterprises treat AI agents as independent identities. Check Point traced a fivefold rise in prompt injection over the same window, and CrowdStrike, Check Point and Trend Micro are now shipping the fix.
macOS 27 is the first release that will not run on an Intel Mac. It arrives with CIOs reporting Apple at 65% of enterprise endpoints and a security window that closes around 2028.
IBM put the global average breach at $4.99 million, up 12% in a year. More than a quarter of malicious attacks were AI-driven, and those added about a million dollars each.
High-risk AI obligations for hiring and workforce management moved from August 2026 to December 2027. The classification problem underneath them has not moved at all.
Splunk found 46% of security teams spend more time maintaining tools than defending, and 52% have thought about leaving cybersecurity altogether. The tooling is the workload.
Regular AI use on corporate devices went from 15% to 45% of employees in a year, and 67% of it runs through personal accounts. The most commonly uploaded asset is company source code.
Malicious open source packages rose 73% in 2025 while PyPI detections fell 43% and NuGet fell 60%. Nearly 90% of everything found was in one registry.
Researchers recovered synced passkey private keys and impersonated privileged Entra ID users despite policies requiring phishing-resistant MFA. The control still works; the marketing around it does not.
NIST expects RSA-2048 and ECC P-256 deprecated by 2030 and quantum-vulnerable algorithms gone by 2035. The obstacle is not the new algorithms, it is producing an inventory of the old ones.
IT teams are losing top engineers to better-paying roles in product and AI startups at an accelerating rate. Leaders who have cracked the retention problem share what actually works.
One survey puts AI agent security incidents at 65% of enterprises, another at 88%. When 82% are still finding agents they did not know about, the disagreement is the finding.
A comprehensive analysis of 300 enterprise workloads finds that the true cost gap between on-premises and cloud is far narrower than most TCO models suggest, especially at scale.
Insurance market tightening is forcing enterprises to demonstrate measurable security posture improvements before renewal. The specific controls carriers are now requiring as a condition of coverage.
New research maps the capabilities separating CIOs who are driving enterprise transformation from those still managing tickets, and the specific behaviours that define the difference.
Broadcom's survey of 1,800 IT decision-makers found public cloud's share of production AI inference fell 15 points in a year. The workload that was supposed to belong to hyperscalers is coming home.
After two years of cloud sprawl, IT leaders are consolidating platforms. Leaders share what the next 18 months of cloud simplification is likely to look like, and which providers will benefit.
A sweeping analysis of security incident data across 500 enterprises asks whether zero trust architecture has delivered on its promise, and what the data reveals about implementation gaps that persist.
Hype aside, what does real enterprise hyperautomation look like right now? We surveyed 250 IT and operations leaders to get an honest read on where deployments truly stand.
Median ransom demands have fallen 65% in two years, yet average recovery costs reached $1.7 million. The money did not leave the ransomware economy, it moved from the extortionist to the clean-up.
Edge deployments have moved from pilot to production across manufacturing, retail, and healthcare. The enterprises that got there first are reporting latency and bandwidth savings that exceed early projections.
CISA's new rulemaking agenda covers areas previously outside its remit. In-house counsel and security officers break down the five areas with the highest near-term compliance exposure for IT organisations.
Security-integrated delivery pipelines have compressed breach-to-detection time significantly at organisations that have made the transition. Here is the framework leading engineering leaders are using.
Quantitative analysis of 800 breach events reveals that the regulatory penalty represents less than 30% of total breach cost, with reputational and operational losses far exceeding initial estimates.
The NIS2 Directive requirements are now in force. A frank assessment of where most organisations stand, which gaps are most urgent to close, and what regulators are likely to scrutinise first.
AI-native ITSM platforms have captured significant market share in three years. We look at which legacy vendors are mounting a credible response, and which are losing ground fast.
New deployment data shows SD-WAN has crossed the mainstream threshold. IT teams that delayed migration are now facing a performance and cost penalty that compounds every quarter.
As AI reshapes every layer of enterprise IT, CIOs face a defining moment: transform the technology operating model or risk being outmaneuvered by faster-moving competitors.
The SEC's final cybersecurity disclosure requirements are in effect. IT and legal teams walk through the six areas most likely to require new data infrastructure and board-level incident reporting.
IT ops teams are deploying AI models faster than governance frameworks can keep up. The organisations that get ahead of this now will have a significant reliability and cost advantage.
New benchmarking data shows that engineering organisations in the top quartile of technical debt management ship features 40% faster than peers, yet most CTOs rank remediation low on their priority list.
Try a different search term or topic filter.
