|
I saw a story recently about ransomware groups still using USBs to deploy ransomware. It's an old method: drop a USB somewhere at random, hope someone plugs it in, and opportunistically deploy something. It goes to show the old methods never really disappear. They just slip out of our minds.
As we get more and more stories about AI being used in insider risk and insider threat, or being part of it, our jobs just get harder. We still need to consider the old vectors, things like USBs, printing and personal email, and make use of the insider risk tools you already have in your business.
We always try to report as much as we can. Hopefully, week on week, these stories really hit home and remind you of all the ways we're faced with risk as professionals.
Headline: Nvidia's CEO Resigns
- Jensen Huang resigned as Nvidia's president and CEO, and from its board, effective immediately. Nvidia announced it on 12 November. CFO Colette Kress is interim CEO.
- It came four days after the board opened an independent investigation into his relationship with Alice Huang, a former Megaspeed executive who US intelligence reportedly flagged as a Chinese intelligence asset. He had admitted to staff he was "unfaithful to his wife".
- Three House committees asked to interview him behind closed doors on 1 December, with a subpoena if he wouldn't come. Committee chair Brian Mast told NBC News that Huang has agreed to brief lawmakers, with the date still being worked out. The Senate Banking Committee wants him on 3 December. The FBI is investigating.
- Nvidia's board says it has "no reason to believe" national security or company secrets were compromised. US intelligence reportedly isn't sure she spied at all. China calls the claims "unsubstantiated".
The media has blown the Alice Huang tie to Beijing out of proportion, with no public evidence. The affair is a different story. Being unfaithful often gets linked to behavioural decisions that could affect your integrity at work. If you're lying to a close partner, you're more likely to lie to colleagues. I'm not sure I fully believe that, but that's the way things are set up now. Everyone lies. Everyone has secrets. It comes down to the person's values, what they believe in most, and which secrets they keep.
Whether stepping down was purely his own decision or the board pushed for it, removing himself from Nvidia right now is the right step. It also shows chip manufacturing and AI are critical infrastructure. We should be treating these companies as national assets and giving them the security support that goes with that.
Source: NVIDIA Announces CEO Transition (Nvidia)Form 8-K, 12 November 2026 (SEC)US intel isn't sure if businesswoman spied on Nvidia for China (CNN)Jensen Huang summoned to testify before Congress after resigning as Nvidia CEO (CNN)
Insider Scoop: The Soldier Inside NATO's Rapid Reaction Corps
- UK counter-terrorism police arrested a 41-year-old US Army member near Gloucester on Tuesday 10 November. Police said the allegations relate to Russia.
- He was based at Imjin Barracks, home of NATO's Allied Rapid Reaction Corps. He is held on suspicion of obtaining or disclosing protected information and assisting a foreign intelligence service, under the UK's National Security Act 2023.
- MI5, the FBI and the US Army worked on the case together. The US military has named him as Lieutenant Colonel Brandon Carroll Tyner. He has not been charged, and is in custody at a London police station.
- The Kremlin says Russia had nothing to do with it.
Not much to say yet. We're still waiting on information to come out. Hopefully in cases like this we can uncover some of the intent or motivation, and see what ties the suspect has with Russia, if that's where this leads.
Source: UPDATED: Member of US army arrested as part of Counter Terrorism Policing investigation (Metropolitan Police)UK-based American soldier suspected of spying for Russia named by US military (PA via Times & Star)American soldier arrested in UK on suspicion of spying for Russia (CBS News)U.S. Army soldier arrested over alleged Russia espionage (CNBC)
Crime Roundup
The assistant who paid the bills.
- Lizbeth Guadalupe Sherman, 47, of El Dorado Hills, California, was an executive assistant to a retired Bay Area venture capitalist. Her job included paying his credit card bills, so she had access to his bank account.
- Between September 2015 and May 2025, she used that access to steal nearly $18 million.
- When investigators searched her home in December 2025, they recovered more than 1,500 luxury items, most in mint condition with the original packaging and tags still on. She also made more than $450,000 reselling items on one platform over nine months in 2025.
- She pleaded guilty to wire fraud on 2 November. She faces forfeiture and restitution of nearly $18 million, and will be sentenced on 22 March 2027.
Source: El Dorado Hills woman pleads guilty to stealing nearly $18 million from her employer (USAO E.D. Cal.)
The broker companies that didn't exist.
- Jared Solomon was a leasing agent at Vornado Realty Trust, a large New York commercial property company, and later its vice president of leasing.
- From 2009 he set up fake broker companies, then claimed broker fees and construction reimbursements for work that was never done. Vornado made 17 payments totalling more than $9.5 million over 14 years.
- He used fake personas, email accounts and a temporary phone number to pose as the fake firms, and forged real people's signatures.
- He spent the money on a $4.55 million house, a $3.6 million Manhattan apartment, a country club membership and a Porsche. He was sentenced on 6 November to 11 years in prison and must forfeit more than $9 million and his interest in both homes.
Source: Former Commercial Real Estate Vice President Sentenced To 11 Years In Prison For Defrauding His Employer Of Over $9.5 Million (USAO S.D.N.Y.)
The insider at the biggest customer.
- Anthony Tepedino founded and ran a New Jersey telecommunications construction company, now defunct.
- Tepedino paid more than $1 million in bribes to a senior manager at the company's biggest customer. In return, the manager steered contracts its way, assigned it work and approved its invoices.
- That customer paid the construction company more than $300 million over about four years. Prosecutors haven't named the customer or the manager.
- Tepedino pleaded guilty on 12 November to conspiracy to commit wire fraud and honest services wire fraud, and aggravated identity theft. Tepedino must forfeit more than $5 million, pay more than $25 million in restitution, and will be sentenced on 17 February 2027.
Source: Former Owner And CEO Of Telecommunications Construction Company Pleads Guilty To Commercial Bribery And Fraud Offenses (USAO S.D.N.Y.)
The takeaway here is trust but verify. Always check the work of people who handle money, contracts and big risk decisions. Don't leave it up to one person and find yourself in the news like these people and organisations.
AI Insider Threat Watch: He Asked the Company's Claude
- Dorsia, a members-only hospitality app, sued its former senior engineer, Chris Mathew, and rival startup Rumor in New York state court on 10 November.
- Dorsia alleges Mathew used his company-paid Claude account to plan his move to Rumor. According to the complaint, he asked how he could leave without legal exposure, what he could take with him, and how to avoid getting caught.
- The complaint says Claude warned him that if he brought "any architectural patterns, customer insights, or pricing intel" to Rumor, "you're toast". Dorsia says he ignored the warning.
- The complaint also alleges Rumor recruited him, encouraged him to share trade secrets and helped hide the arrangement. Both defendants reject the claims. Rumor's lawyer called the suit "devoid of merit".
If the company can show the transcripts and prove he was the one writing the prompts, this could be a pretty big case of a company using collected prompts as proof. There's still the traditional forensic work, though. They need to prove he was at the keyboard and the one prompting the AI for this to stand up.
Source: Elomi Corporation d/b/a Dorsia v. Rumor Group Corporation and Chris Mathew, Index No. 656375/2026 (NY Supreme Court, New York County; search NYSCEF by index number)App Says Ex-Engineer Stole Secrets, Used AI To Plan Move (Law360)Dorsia Says Engineer Asked Its Own Claude Account How to Take Company Secrets Without Getting Caught (Tech Times)
Mitigation Corner: The Logins Nobody Turned Off
The marketing account that stayed open.
- A.W.G. Dewar, a US insurance agency, and its affiliate Intact Insurance Group USA have sued a former employee in federal court in Massachusetts. The complaint was filed in September.
- Dewar says it fired her on 17 April 2026 and shut off her internal logins. Her login to Constant Contact, the email marketing platform, stayed active.
- Five days later, on 22 April, she allegedly logged in and exported the names of current and prospective client schools, with contact names, titles and email addresses. Dewar says she joined a competitor, Vertical Insure, and has since approached its customers.
The AI note-taker on a personal Gmail.
- In October, benefits brokerage Sequoia sued a departing broker in federal court in Colorado.
- It alleges she used Granola, an unapproved AI note-taking app linked to her personal Gmail account, to capture 18 sets of detailed client meeting notes, and kept them after joining a competitor.
Both cases are allegations, and neither has been tested in court. Apple's case against OpenAI has a similar thread. OpenAI argued that former Apple staff kept access because of weaknesses in Apple's offboarding, not because they meant to keep anything.
Seeing these cases tested in court will be really interesting. I think both are failures on the employer's side, though: not controlling the accounts people still had access to once they'd left. This one is on the employer. They should have better offboarding checklists, and better control of what applications are used during people's employment, including non-approved AI tools.
Make offboarding a checklist of every system, including the SaaS tools and AI apps nobody thinks of, and review what the person touched in their last 30 days.
Source: Ex-staffer's post-termination login allegedly gave competitor a 33-year client list (Insurance Business)Insurer sues former employee over alleged customer database theft (Claims Pages)Sequoia sues after departing broker allegedly used AI app to grab client data (Insurance Business)
Trusted Profession Watch: The Judge Who Cited Cases That Don't Exist
- Lawrence Wheeler is a judge in Stephens County, Oklahoma. In a custody case last November, he refused one parent's request for the other parent to have a psychological evaluation, and cited earlier Oklahoma cases to back it up.
- At least two of those cases don't exist. Wheeler admitted to Oklahoma State Bureau of Investigation agents that they came from ChatGPT. He says he wrote the order himself and only used AI for "research".
- The Oklahoma Council on Judicial Complaints asked the bureau to investigate. The Attorney General's office said the evidence doesn't support a criminal prosecution. Any discipline is up to the Oklahoma Supreme Court.
- The mother in the case, Adriane Capers, says it has been running for about three years, mostly in front of Wheeler.
There should be harsher rules and penalties for professionals who use AI and let it hallucinate without checking. People need to be held accountable for not doing their job.
Source: Oklahoma judge admitted to citing fake ChatGPT cases in court order, investigators say (KFOR)Stephens County mother speaks out after judge accused of using ChatGPT in custody case (News9)
Trivia
This week's Headline is about Nvidia. In 1993, Jensen Huang and his two co-founders planned the company in a booth that now has a plaque on it. Where is the booth?
A) A Stanford University library B) A Denny's restaurant in San Jose C) A garage in Palo Alto D) A Taipei tea house
Corporate News: The Finance Director Who Knew the Numbers First
- Jesse Mitchell was a senior director in The Trade Desk's financial planning and analysis team, from around June 2024. He could see revenue and earnings results before they were made public.
- In August 2024, during a blackout period, he bought The Trade Desk shares ahead of its second-quarter results, making about $19,700.
- In February 2025, The Trade Desk missed its revenue forecast for the first time since going public in 2016, and its shares fell about 30%. Mitchell had bought put options, which pay out when a share price falls, and made about $318,000.
- He pleaded guilty to securities fraud on 9 November and agreed to forfeit $338,058.56. He faces up to 20 years and will be sentenced on 8 March 2027.
Someone trading on insider information. I wonder who else is doing this.
Source: Finance Director Pleads Guilty to Insider Trading (USAO S.D.N.Y.)Jesse R. Mitchell, Litigation Release No. 26614 (SEC)
Around the World
China: the colleague who lost the posting.
- China's Ministry of State Security published a case on 10 November about a state employee it calls Liu. In 2024, Liu lost an overseas posting to a colleague, Li.
- Liu then sent an anonymous letter to a foreign organisation, falsely claiming Li was a spy.
- The ministry said Liu only wanted to vent a personal grudge and didn't mean to harm national security, but the false report still damaged it. Liu got an administrative warning under China's counter-espionage rules.
- The ministry didn't name Liu's employer or the foreign organisation, or say how it found out.
Source: Ministry of State Security case report, 10 November (Sina, in Chinese)China warns against fake spy claims after jealous public servant frames rival (SCMP)
Japan: the retired engineer and the hard drive.
- On 27 October, prosecutors in Nagoya indicted Shoichi Furukawa, 66, a former Asahi Kasei employee, for disclosing trade secrets under Japan's Unfair Competition Prevention Act.
- According to reports of the indictment, he took a hard drive of research and manufacturing data when he left the company in 2018. In August and September 2024, he allegedly emailed files with trade secrets to people at a Chinese company, which Asahi Kasei names as Shandong Shengquan New Materials. A company set up by his wife had signed a consulting deal with the Chinese firm that May.
- He reportedly told investigators he wanted to show the Chinese company's staff how much he knew, and to keep his standing as a well-paid expert. Police arrested him again in October over a second suspected leak, about a coating used to make semiconductors.
- Asahi Kasei says it is reviewing access rights and tightening control of laptops lent to staff who are about to leave. It plans to sue him and the Chinese company.
Source: 旭化成情報漏洩事件、元社員「地位保とうとした」 名古屋地検が起訴 (Asahi Shimbun via Yahoo! Japan News)Statement on arrest of former employee, 17 September 2026 (Asahi Kasei)
Here we start to get into news from around the world, as I search sources other than English ones for insider threat and insider risk. Hopefully this section grows as the weeks go on.
What's On
iTA Workshop: Bring Your Own Model: The Insider Risk of Unmanaged AI, a DTEX Insider Threat Advisory workshop.
- When: Tuesday 8 December 2026, 1:00 PM to 2:00 PM PT (Wednesday 9 December, 7:00 AM AEST Brisbane time).
- What it covers: the insider risk when staff run AI agents on unapproved or unrestricted models, outside corporate guardrails.
- Register: https://www2.dtex.ai/l/464342/2026-11-09/6rptxq
AI-Driven Insider Threat, a FINRA webinar for people at FINRA member firms.
- When: Wednesday 18 November 2026, 2:00 PM to 3:00 PM ET (Thursday 19 November, 5:00 AM AEST Brisbane time).
- What it covers: how opaque, connected AI systems, including AI built into vendor products, can create insider-like risk without anyone meaning to.
- Online. Member firms only: you need a CRD or State Regulator number to register.
- FINRA follows it with a team decision simulation, "Trusted Access, Unintended Consequences", on Wednesday 2 December, 2:00 PM ET (Thursday 3 December, 5:00 AM AEST). Teams work an AI-driven insider incident with incomplete information.
- Register: https://www.finra.org/events-training/conferences-events/ai-driven-threats-and-consequences
Still coming up from last week: SPRITE+ "From oops to ops" (Wednesday 18 November, 1:00 PM GMT, 11:00 PM AEST) and Charles Finfrock at SecureWorld East (Wednesday 18 November, 10:00 AM ET, Thursday 1:00 AM AEST). The Adelaide networking drinks are on 7 December, and the GovInfoSecurity webinar with Ent is on 10 December.
On the horizon: Carnegie Mellon SEI's Insider Threat Program Manager course runs live online on 20 to 22 January 2027, 8:30 AM to 4:30 PM ET (11:30 PM to 7:30 AM AEST overnight, so a tough one from Australia). From US$2,250. Register by email: course-info@sei.cmu.edu. https://www.sei.cmu.edu/training/insider-threat-program-manager-implementation-operation . The Australian Insider Risk Symposium in Sydney is still "November, date TBD" with registration closed.
Resources
A few places worth your time if you want to go deeper on insider risk.
Worth bookmarking:
And my own: Insider Threat Awareness Training on YouTube, a free course with certification.
Trivia Answer
B. A Denny's restaurant on Berryessa Road in East San Jose. Jensen Huang, Chris Malachowsky and Curtis Priem planned Nvidia in a booth there in 1993. Denny's put a plaque on the booth in 2023.
Source: Nvidia was born in Denny's, and now a booth in the East San Jose restaurant has been dedicated to the trillion-dollar company (PC Gamer)
|