AI at work. Fake Teams installers. Meta to start using AI interactions to target adverts.
TikTok's importance in US news consumption. More high-profile security incidents. And the internet as cargo cult.
Ubiquitous
The year began with AI dominating the Consumer Electronics Show in Las Vegas, where the theme was AI everywhere all the time. This week that was exactly the message about AI in the workplace. As consulting behemoth Accenture put it in an earnings call, “advanced AI is becoming a part of everything we do.” Over 550,000 reinventors (that’s Accenture-speak for employees) have been trained in the fundamentals of Generative AI, it said. Anyone incapable of that “upskilling” is being “exited on a compressed timeline.” Indeed 11,000 reinventors have already been reinvented out the door.
Elsewhere, Microsoft announced that’s it’s adding AI agents to Word, Excel and Powerpoint to help Copilot users create, edit, and analyse documents, spreadsheets, and presentations. The aim is to enable even basic users to take advantage of advanced functions. As an example Microsoft says you could ask Excel to analyse a set of sales data and ask it to provide insights to help make decisions about the business.
Organisations need a specific subscription in order to use this functionality but no problem if your workplace doesn’t have it; Microsoft says you can just use your personal account instead. This seems pretty extraordinary, given that it amounts to sanctioning the use of unapproved (or “shadow”) IT at work. But it’s fine, Microsoft assures us, because IT managers retain “full control and oversight” and “enterprise-grade data protection is maintained.” We’re sure IT departments are delighted by the prospect of managing all this.
Threats
Teams: Attackers are exploiting search results to promote fake Microsoft Teams installers that infect Windows devices with malicious software that provides initial access to corporate networks. The adverts lead to a (convincing) website impersonating the real Microsoft one. Bleeping Computer
Insiders: You may have seen the story about hopeful hackers trying to persuade the BBC’s cybersecurity reporter to hand over his network credentials. “You’ll never need to work again,” was their pitch. Unsurprisingly, he refused. We wish we could say this type of offer is uncommon but, alas, it’s not and should be on every organisation’s risk register.
Supervisor: Here’s a good example of the tactics employed by phone scammers to make their calls more credible. A New York Times crime reporter explains how he very, very nearly fell for a scam despite spending decades writing about them. In this case, what nearly caught him out was when the caller (who claimed to be from a bank) transferred him to his “supervisor.”
PDFs: Another reason to be cautious about opening PDF files. Varonis has spotted a new toolkit that allows attackers to convert ordinary PDF files into interactive lures designed to bypass email security.
Facebook: Scumbags are using Facebook groups that promote social activities for the elderly to trick victims into installing malicious software on Android devices. The campaign was first seen in Australia but has now spread globally. ThreatFabric
Fuel: Elderly people in the UK are also the target for text messages urging them to sign up for winter fuel payments. The messages are relatively realistic and include a web address that at first sight looks like it might be a genuine government one.
Messaging: Two new spyware campaigns are targeting Android users with fake upgrades or plugins for the Signal and ToTok messaging apps to steal sensitive data. ESET says that currently the campaigns are focused on the UAE.
AI
Listening: Meta says it will start using people’s conversations and interactions with its AI services to create personalised content and advertising. “For example, if you chat with Meta AI about hiking, we may learn that you’re interested in hiking – just as we would if you posted a reel about hiking or liked a hiking-related page,” Meta explained. On the same day, and possibly not coincidentally, the head of Instagram posted a “myth-busting” video insisting it doesn’t listen to what people are saying. “We do not use the phone’s microphone to eavesdrop on you.” This “would be a gross violation of privacy” and in any case would drain your phone’s battery, Adam Mosseri said. While we’re sure Meta hasn’t been listening to what we’re saying (not least because they don’t need to), effectively that’s exactly what they are going to do once they start analysing AI usage on December 16.
Cameo: OpenAI has a (frankly alarming) new version of its Sora app which adds sound and gives you “the power to step into any world or scene and let your friends cast you in theirs.” Open AI says Cameo is about “creativity and joy.” It’s undeniably impressive but we suspect it will also be about fraud and disinformation. Or maybe we’re just getting cynical.
Medicine: Current medical AI models may perform well against standard benchmarks but the scores don’t mean they can be trusted for reliable medical reasoning. That’s the finding of Microsoft researchers who say the “benchmarks reward test-taking tricks over medical understanding.”
Hacks: Video deepfakes of broadcast journalists around the world are being used to promote products online. NiemanLab has case studies illustrating how the likenesses of hacks are being hijacked.
Confession: A college student in Missouri has discovered that telling ChatGPT he’d vandalised 17 vehicles and asking it whether he would go to jail is a somewhat flawed idea. When police searched his phone, not only did they find location data putting him at the scene of the alleged crime, but also the rambling conversation with ChatGPT in which he inquired if there were any cameras in the car park he had just left. Ozarks First
Trump
TikTok: One in five people in the US say they get their news regularly from TikTok, according to Pew Research Center analysis. That’s up from 3% in 2020 and, Pew adds, it’s the fastest growth in news consumption for any social media platform it has studied. These figures come as a group of Trump-supporting oligarchs is about to take control of TikTok. For his part, President Trump said the platform should be tweaked to become “100% MAGA,” while adding that he was only joking.
Tracking: US Immigration and Customs Enforcement (ICE) has bought access to a surveillance tool that harvests billions of pieces of location data from hundreds of millions of mobile phones, according to documents reviewed ($) by 404 Media. The data provides extraordinarily granular information about the movements and activities of mobile phone users. Meanwhile, the Electronic Frontier Foundation spent a week at a US police convention; its key conclusion was that “the name of the game now is consolidating as much information as humanely possible from surveillance devices, the internet, other governmental data, and literally a million other places.” Security veteran Bruce Schneier says this type of surveillance requires everyone in the US to reconsider their personal “threat model,” i.e. who you are, what you have, who might want it - and how they might use it against you.
Blocked: As Illinois governor J D Pritzker claimed President Trump is suffering from dementia, Google is reported to be blocking AI search summaries for the query “does Trump show signs of dementia.” This doesn’t appear to apply to similar searches about other presidents. The Verge
In brief
Out of control: Another week, another set of high profile security incidents. Among the latest are Japanese brewing giant Asahi, luxury retailer Harrods, and reportedly the US Air Force. Gartner says cybersecurity leaders must stop seeing resilience as a “tick box exercise” if they’re to achieve meaningful protection.
JLR: On that subject, researchers say they warned Jaguar Land Rover in June that hundreds of credentials associated with its employees were circulating on the dark web. The researchers say they urged JLR to get in touch but didn’t receive a reply to their messages. Bloomberg
Order: The UK government is having another go at forcing Apple to give it access to user data in its cloud storage service, according to sources quoted by the Financial Times. A previous attempt would have applied to all data and appears to have been scuppered by the US. This time it’s restricted to UK users. Meanwhile, Signal says it will withdraw its secure messaging app from the EU if it passes proposed Chat Control legislation.
Takedown: Interpol says 260 people suspected of involvement in online romance and extortion scams have been arrested in Africa. An operation took place in July and August across 14 countries. The scams had targeted 1,400 victims who lost nearly $2.8 million, Interpol said.
Teens: Two 17-year-old Dutch boys have been arrested on suspicion of carrying out espionage on behalf of Russian hackers. They are alleged to have walked past Europol, Eurojust, and the Canadian Embassy using equipment designed to harvest information about WiFi networks. Reports say they were approached on Telegram by a pro-Russian hacker. NL Times
Sendit: The US Federal Trade Commission has opened legal proceedings against the Sendit anonymous messaging app. The FTC says Sendit used fake accounts to message users and trick them into paying a subscription to see the sender’s message. Sendit is also accused of failing to obtain parental consent before collecting the data of children under the age of 13.
Tile: We were early adopters of Tile Bluetooth trackers, so it’s depressing to learn they leak identifying data in plaintext, according to research from Georgia Tech researchers. They found a number of flaws that they say undermine the manufacturer’s security and privacy guarantees. The Register
Illusions: The internet is a cargo cult filled with plausible sounding “BS”, according to a Substack essay by Dave Shapiro. If you haven’t come across him, Shapiro describes himself as someone “focusing on the future of humanity as we navigate the Fourth Industrial Revolution.” He says he’s now decided to slowly step back from the internet (which is what made him famous) to focus on more substance in the offline world.
Updates
Apple: Updates for iOS, iPad and macOS devices. These include an important security fix to address a vulnerability that could be exploited by using malicious fonts. New versions of iOS 26 also fix some early problems.
Google: A new feature in the Drive app is designed to detect and combat ransomware attacks. It will use AI to detect infections and stop them being spread to the cloud. Google has also announced that Gmail enterprise users can now send end-to-end encrypted emails to people on any email service or platform.
Outlook: Microsoft has published a workaround for an an issue that triggers errors when opening encrypted emails sent from other organisations. It involves making configuration changes that are outlined here.
Windows 11: 25H2 is a minor update which, according to Microsoft, includes better vulnerability detection. The update is gradually rolling out worldwide.
Cisco: The UK and US cybersecurity agencies have urged organisations to get rid of end-of-life Cisco security appliances. The Adaptive Security Appliance devices have two vulnerabilities that are being actively exploited, and monitoring suggests there are still some 50,000 of them in use. Any organisations still using these appliances should apply updates - and if they can’t they should stop using them.
MyCloud: Firmware updates for multiple My Cloud network-attached storage devices to address a critical-severity vulnerability that could be exploited remotely to execute arbitrary system commands.
VMWare: Fixes for six vulnerabilities affecting VMware Aria Operations, NSX, vCenter, and VMware Tools products, including four high-severity flaws.
Adobe: A Premiere Pro app has been released for the iPhone. For the moment, it’s free to use - although payment will be required to use some AI features.