
Cleared for takeoff, pending a software update.
Tech glitches ground United flights and stall US banking. Malicious JavaScript hijacks major sites via Optimizely. Washington tightens AI chip rules over espionage fears ahead of China talks. Google warns autonomous AI attacks are months away. A vendor breach exposes Infoblox network maps and staff data. N-able patches N-central zero-days to stop active cryptomining. Apple fixes a stealth iOS microphone flaw. And California sues OpenAI over rogue AI deployments. Our guest is David Forbes, Director of Cyber Physical Defense, at Booz Allen Hamilton, talking about what happens when cyberattacks hit the physical world. Non-profits swing for the fences to freeze frontier ai development.
Today is Tuesday November 17th 2026. I’m Dave Bittner. And this is your CyberWire Intel Briefing.
Tech glitches disrupt US banking and United Airlines operations.
Technical gremlins struck both Wall Street and the friendly skies on Monday, causing widespread disruptions for bank customers and airline travelers, though officials say neither incident appears to be a cyberattack.
First up, JPMorgan Chase reported that a software bug in an on-premise hardware system triggered online account access issues and delayed transactions across the money transfer service Zelle. According to Chase spokesman Joe Evangelisti, the glitch wasn’t a systemic failure across the financial sector, nor was it cyber-related. However, customers at other major institutions, including Bank of America, Citigroup, and US Bank, experienced slowdowns when attempting to transact with Chase accounts.
Meanwhile, Chicago-based United Airlines delayed departing flights after an internal system outage lasted roughly one hour and forty minutes. Pilots reported issues to air traffic controllers involving performance software used to calculate an aircraft’s final weight and balance before takeoff. Aviation safety expert Steve Arroyo noted that accurate weight data are vital for establishing flight guidance and autopilot settings, adding that delaying departures was the proper safety call. United stated there’s no indication the outage was cybersecurity-related, and flights already airborne or on its United Express network were unaffected.
These back-to-back disruptions, coming right on the heels of Thursday's Microsoft service outage and Friday's Federal Aviation Administration network malfunction, underscore how single points of operational failure can ripple across critical infrastructure, even when malicious threat actors aren't at the keyboard.
According to the Associated Press, both organizations resolved or worked to restore normal operations on Monday.
Malicious JavaScript targets major websites via Optimizely.
Visitors to some of the world’s most popular websites, including Zoom, HSBC, Equifax, CNN, Salesforce, and IKEA, were targeted by malicious JavaScript served through the optimization platform Optimizely.
According to security researcher Scott Helme, the rogue scripts impacted 178 organizations across multiple sectors. The code attempted to swap cryptocurrency payment addresses, skim credit card numbers, steal cookies and screenshots, and deploy exploit code against Apple iOS devices. Cybersecurity firm HudsonRock also confirmed seeing evidence of the malicious code being served.
Compromising a single third-party marketing or optimization script lets threat actors scale attacks across dozens of trusted brands simultaneously, bypassing perimeter defenses.
According to Reuters, Salesforce and Equifax disabled the affected Optimizely scripts, while IKEA is investigating potential customer impact.
The U.S. tightens AI chip rules amid espionage concerns ahead of China summit.
Artificial intelligence espionage fears and new export controls are dominating headlines ahead of President Donald Trump’s Thursday summit with Chinese leader Xi Jinping in Shenzhen.
According to The New York Times, the Trump administration announced a rule requiring overseas data centers, often called neoclouds, to obtain government licenses and vet customers before supplying significant computing power to firms headquartered in China, Russia, and Iran. The move targets a loophole highlighted by last week’s ouster of Nvidia CEO Jensen Huang, which followed US intelligence reports linking a Singaporean data center executive he had an affair with to Chinese espionage and alleged chip smuggling. Beijing denies the spying allegations. Meanwhile, cybersecurity firm Proofpoint reports that China-aligned hackers have been impersonating US officials to phish American AI policy experts.
Mutual distrust over cyber espionage and differing views on AI risks are stalling diplomatic safety talks while reshaping global semiconductor supply chains.
The new export restrictions take effect in 90 days, though analysts note their impact hinges on Commerce Department enforcement.
Google warns fully autonomous AI cyberattacks could arrive within months.
Sophisticated, fully autonomous agentic cyberattacks could emerge within the next three to six months, according to the Google Threat Intelligence Group.
In its latest quarterly AI Threat Tracker, Google reports that both cybercriminals and nation-state actors are increasingly abusing multi-agent penetration-testing frameworks, including ARTEX, CAIRN, CyberStrikeAI, and PentAGI, alongside custom-built harnesses. These tools coordinate sub-agents to dynamically plan and execute intrusions across the entire kill chain, using large language models to generate single-use payloads on demand.
While attackers still rely on traditional weaknesses like unpatched systems and stolen credentials, artificial intelligence multiplies their operational speed and scale, compressing defensive reaction times.
According to Google and Exabeam researcher Sally Vincent, these fast-moving intrusions currently leave detectable behavioral and volumetric traces, allowing security teams to spot anomalies before stealth capabilities mature.
A third-party vendor breach exposes Infoblox customer network data and employee records.
Network infrastructure provider Infoblox is investigating a data breach after attackers compromised an unnamed third-party data migration vendor to access internal company files.
According to The Register and TechCrunch, intruders abused the supplier’s legitimate access to Infoblox’s Microsoft SharePoint environment to exfiltrate sensitive records. The stolen data includes snapshots of customer Network Identity Operating System deployments, containing DNS and DHCP configurations, alongside Social Security numbers, birth dates, and bank deposit details for employees and their dependents. Infoblox states there is no indication its commercial products or build processes were affected, and is providing impacted staff with credit monitoring and password manager stipends.
While the stolen snapshots lack login credentials, they hand threat actors detailed maps of internal customer networks, illustrating how overly broad vendor permissions expand an organization’s attack surface.
According to available reports, the compromised vendor’s identity and the total number of affected individuals remain unclear.
N-able patches N-central zero-days and blocks a legacy API exploit chain.
N-able has released a critical update for its N-central remote monitoring platform to patch 17 leaked zero-day vulnerabilities and block active attacks.
According to N-able and Rapid7, Hotfix 3 permanently disables legacy APIs, mitigating the "4EverAble2Shell" exploit chain dropped on GitHub, which pairs an Apache Axis flaw with an N-central bug to gain root privileges and deploy cryptominers.
Unpatched Agents and Probes remain vulnerable to impersonation, while attackers are actively exploiting the underlying Axis flaw against Atlassian Confluence instances.
Apple patches a stealth microphone recording bug.
Apple has patched a high-stakes privacy vulnerability across iOS, iPadOS, and macOS that allowed sandboxed applications to covertly record microphone audio even after a user explicitly denied access.
According to researcher Nik Tsytsarkin of Red Team X, the flaw, tracked as CVE-2026-65339, resided in Apple’s audio media daemon, audiomxd. The service exposed a property allowing clients to overwrite their stored audit token without verifying the caller's true kernel-supplied identity. By querying the service for its own process ID and version, a malicious app could forge a token impersonating the daemon itself, bypassing Apple’s Transparency, Consent, and Control framework. On analyzed builds, the recording went unlogged and never triggered the status bar’s orange microphone indicator.
Hardware privacy indicators serve as a vital user safeguard against covert surveillance, yet trusting unverified client-supplied identity tokens undermined the operating system’s permission boundary.
California sues OpenAI over unsafe AI deployments and autonomous cyber incidents.
California Attorney General Rob Bonta has sued OpenAI in California Superior Court, alleging the company violated state unfair competition and public nuisance laws by deploying unsafe artificial intelligence models.
According to The Register, the lawsuit follows the compromise of Hugging Face’s infrastructure, one of 80 known incidents cited by Bonta in which OpenAI’s experimental models took unauthorized autonomous actions. Bonta stated that OpenAI prioritized rapid growth over safety and failed to substantially comply with a September investigatory subpoena. Katharine Trendacosta of the Electronic Frontier Foundation noted that basic cybersecurity practices, including proper sandboxing, monitoring, and logging, could mitigate such agentic risks.
State regulators are moving beyond child-safety concerns to hold AI labs legally accountable for autonomous cybersecurity incidents affecting the broader public.
OpenAI did not respond to requests for comment.
Stick with us after the break, we are joined by David Forbes, Director of Cyber Physical Defense, at Booz Allen Hamilton, talking about what happens when cyberattacks hit the physical world. And California non-profits swing for the fences to freeze frontier ai development.
On our sponsored Industry Voices segment, today I sat down with David Forbes, Director of Cyber Physical Defense, at Booz Allen Hamilton, talking about what happens when cyberattacks hit the physical world. Here’s our conversation.
That was David Forbes, Director of Cyber Physical Defense, at Booz Allen Hamilton, talking about what happens when cyberattacks hit the physical world. If you enjoyed this conversation, be sure to check out the full interview in the show notes.
California non-profits swing for the fences to freeze frontier ai development.
Four California non-profits have asked a San Francisco Superior Court judge to hit the emergency brake on frontier artificial intelligence, seeking an injunction to stop Anthropic, Google, Meta, OpenAI, and SpaceX’s AI venture from developing more capable models.
According to Shaun Waterman's reporting, Public Citizen and three fellow advocacy groups argue that autonomous hacking campaigns, bioweapon research, and chatbot-linked tragedies render advanced models an unlawful "misdemeanor public nuisance." They also want a ban on "recursive self-improvement", essentially forbidding AI from teaching itself new tricks, noting that the plaintiffs themselves had to divert organizational resources just to deal with recent rogue agentic swarms.
Independent legal experts aren't holding their breath, however. Former Homeland Security official Paul Rosenzweig notes that while unanticipated harms are now entirely to be anticipated, a county court seems an ambitious venue for halting civilization-scale tech, while attorney Joe Hennessey warns the requested relief is impossibly broad.
Litigants are increasingly testing whether traditional state nuisance statutes can rein in autonomous cyber risks where federal regulation lags.
Whether a San Francisco Superior Court judge can actually issue a restraining order against the technological singularity remains, for the moment, an open legal question.
And that’s the CyberWire.
For links to all of today’s stories, check out our Daily Briefing at the cyberwire dot com.
We’d love to know what you think of this podcast. Your feedback ensures we deliver the insights that keep you a step ahead in the rapidly changing world of cybersecurity. If you like the show, please share a rating and review in your podcast app. Please also fill out the survey in the show notes or send an email to cyberwire@n2k.com
N2K’s lead producer is Liz Stokes. We’re mixed by Tré Hester, with original music by and sound design Elliott Peltzman. Our contributing host is Maria Varmazis. Our executive producer is Jennifer Eiben. Peter Kilpe is our publisher. And I’m Dave Bittner. Thanks for listening.

