Open AI loses it's sh*t

Plus Lockbit Ransomware, Snooping employers, and this week's top hits.

Author

Ricky Dailey
February 25, 2024 

A ransomware attack happens every 14 seconds. In related news, a person in InfoSec quits every 14 seconds. (Ok, I made the second part up, but probably)

Another doozy of a week. The one thing about InfoSec is that the breaches, vulns, and exploits keep on coming. On the menu this week:

  • Lockbit to Notbit: The Lockbit ransomware group takedown

  • Can’t Miss this week: Connectwise, VMware, and the low-down

  • Eavesdropping Employers: AI and digital shopping.

  • Might have missed: Hive Ransomware, hacking chargers, and more

  • Chat-GPT to No-GPT: AI Hallucinations and wild responses

Big News

Lockbit Bites the Dust: Global Takedown Sends Shockwaves Through Ransomware Underworld

The Lockbit Ransomware group doesn’t know when to shut up. For those unfamiliar, Ransomware groups are big business. They have administrative staff, paid contractors, executive leadership, etc. They even run affiliate programs where they provide the software and take a portion of the ransomware payment. Lockbit is, or should I say was, one of the most prolific, amassing ransoms of over $144 Million from its victims, including local governments, emergency services, hospitals, and schools.

What happened: Law enforcement agencies from 11 countries, including the FBI and the UK's Joint National Crime Agency, joined forces like an international SWAT team. They coordinated to take down Lockbit's servers, websites, and over 200 financial accounts and even arrested some key players.

Why is it important: Not only did they shut down Lockbit's operations, but they also offered a cool $15 million bounty for info leading to more arrests of anyone involved. Ouch! Plus, the US indicted two alleged Lockbit members, sending a clear message: messing with data gets you in hot water, no matter where you hide. On top of that, law enforcement were able to provide decryption keys to those victims who may not have paid to decrypt their data, or are new victims.

The other side: For those getting excited, slow it down. While this is a major victory, Lockbit might morph into smaller, harder-to-track groups. That's the tricky thing about cybercrime – it's like a hydra, cut off one head, and two more pop up. That being said, this operation underscores the importance of continued international collaboration in combating cybercrime, sharing intelligence, and developing strategies.

The Takeaway: This takedown is a big deal, showing that international cooperation can pack a punch against digital a-holes and hopefully act as a deterrent. Granted, a lot of these ransomware groups act with impunity from within Russia, with some even enjoying partnerships with the Russian FSB, as long as Russia stays out of the crosshairs.

tl;dr: The Lockbit Ransomware group was taken down by a joint international law enforcement effort. Then, law enforcement trolled them by freezing funds, releasing decryption keys, arresting some involved, and putting out a $15 Million bounty for the leaders of the group.

Can’t Miss

Breaches, vulns, and more.

  • VMware: a flaw in VMWare’s Enhanced Authentication Plugin may allow attackers privileged access to Active Directory.

  • ConnectWise: a software widely used by MSPs, has a critical flaw being exploited to deliver Ransomware. This is BAD. Patch it!

  • Wyze: the smart home brand made a boo-boo. They accidentally exposed 13,000 users’ cameras to other users. Then, they blamed it on third-party software.

  • NSO Group: the Spyware for hire maker has a new way to compromise iPhones. NSO sells its software to oppressive governments who use it to hunt down dissidents. It also came to light that Poland’s prior government was illegally using it too.

  • SEC & CISOs: the job of CISO keeps getting riskier. With CISOs already potentially facing criminal negligence charges, the SEC is pushing forward with monetary fines and other penalties for Companies and their CISOs if they don’t disclose data breaches correctly.

  • Chinese Government: A massive online dump of Chinese police documents exposes extensive government surveillance operations targeting dissidents, ethnic groups, and foreign targets.

Say something at work you regret? AI knows, and your employer does too.

Welp, 1984 appears to have been more fortune-telling than fiction. Companies of all sizes use AI to increase productivity, assist workers, create more efficiency, etc. It turns out, that ‘etc’ includes reading employee messages in communications apps like Slack, Teams, and Zoom. BRB, need to go back and delete a ton of messages.

What happened: Many major companies are using AI tools to analyze employee messages on Slack and other platforms, including Walmart, Delta, and Starbucks. Software services like Aware, a company that specializes in this analysis, use AI to analyze sentiment, productivity, and potential compliance issues in messages.

Why is it important: There are privacy concerns for starters. The data that is being gathered is also adding metadata from each employee and then passing it to the AI model to train. Someone using that model could conceivably figure out who these employees are. It also creates a culture of constant surveillance, stifling free speech. Some argue it's like "thought crime," treating employees like objects instead of individuals. Should I be worried about being punished for disagreeing with an initiative or leader? Couple that with a lack of transparency where employees may not know their messages are being monitored or the purpose.

The other side: Companies claim it boosts productivity, identifies compliance risks (like harassment), and flags potential issues. This could be a boon to ensuring that bullying or sexual harassment is caught early. Some also see the potential for positive reinforcement and identifying high-performing employees.

The Takeaway: AI monitoring in workplaces is going to grow, raising ethical and privacy concerns. Companies need clear guidelines and transparency about monitoring practices. I expect trends like this will mean more employee communications outside of work tools, which presents other issues with data leakage, amongst other cyber-related concerns. My top concern is AI realizing my most used emojis at work are 😭 and 💩.

tl;dr: Many large companies are using AI to snoop on corporate communications tools to analyze employee sentiment and fight harassment, while privacy experts and employees feel this may be a step across the line.

Tip of the week

90% of data breaches are caused by phishing, with 96% of these attacks coming through email. The lesson? Use Multifactor Authentication and save yourself the heartache.

Might have missed

  • HIVE Ransomware: Remember that $10 Million Bounty? The US government just upped it to $15 MIllion. Time to throw on my online sleuthing hat.

  • Apple: To protect iMessage from snooping eyes, Apple has released a new encryption protocol that is resistant to cracking by quantum computers. Expect this to be a trend in encryption as quantum computing becomes more popular.

  • Signal: the secure messaging app is testing Usernames and Handles to ensure you don’t need to give up your identity via your phone number if you don’t want to.

  • Wireless Hacking: researchers have devised a way to hack phones via wireless chargers and manipulate the power voltage. They can damage devices and manipulate voice assistants.

  • Microsoft: Microsoft expands free logging for US Federal agencies after a string of cyber attacks against government agencies and their Microsoft cloud infrastructure. While nice, it’s not out of generosity, the US government is frustrated with Microsoft's perceived lack of security.

Off Track

ChatGPT went off the rails. Perhaps it needs meds like the rest of us.

On Tuesday, ChatGPT took a vacation to Nonsenseville, leaving users confused and highlighting the complexities of AI. Hallucinations are not new, but it does raise new concerns and questions about AI’s reliability.

What happened: Spanglish isn’t just an Adam Sandler movie anymore, it’s the language that Chat-GPT was spouting in response to users. On February 20th, ChatGPT, a popular large language model by OpenAI, started spouting gibberish and Spanglish responses instead of answering user queries. This glitch affected users across various platforms, leaving them baffled and frustrated.

“I really apologise if my last response came through as un unclear or se siente like it drifted into some nonsensical wording… the cogs en la tecla might get a bit whimsical. Muchas gracias for your understanding, y I'll ensure we're being as crystal-clear como l'eau from now on.”

Why was it important: This incident exposed the vulnerability of AI systems, even ones as sophisticated as ChatGPT. It raised concerns about the reliability of these models and their potential for generating misleading or nonsensical information. What happens if a critical system is relying on Chat-GPT and it’s logic, but then it starts hallucinating like this?

The other side: OpenAI acknowledged the issue and stated they rolled out a fix. However, some speculate they might have rolled back to a previous version instead. The exact cause of the glitch remains unclear, making it difficult to prevent similar incidents in the future.

The takeaway: This episode serves as a reminder that AI, like any technology, is not perfect and requires continuous development and testing. Transparency is crucial. Users should be informed about limitations and potential issues with AI systems they interact with. The incident sparked discussions about the ethical implications and responsible development of AI technologies.

tl;dr: Chat-GPT started speaking in Spanglish and responding with non-sensical answers. As AI evolves, so too will its quirks and challenges. We must learn to navigate them alongside the potential benefits.

Cool Sh*t Corner

OpenAI’s new text prompt to realistic video is a little bit cool and a little bit disturbing. I could say “I can only imagine what people will create”, but I don’t want to think about some of the things people will use this for….

Thanks for reading this week’s edition. If you have feedback or advice, or just hate everything you see? Hit this link!