Cyber Intelligence Weekly (December 3, 2023): Our Take on Three Things You Need to Know // Echelon Risk + Cyber

Welcome to our weekly newsletter where we share some of the major developments on the future of cybersecurity that you need to know about. Make sure to follow my LinkedIn page as well as Echelon’s LinkedIn page to receive updates on the future of cybersecurity!

To receive these and other curated updates to your inbox on a regular basis, please sign up for our email list here: https://echeloncyber.com/ciw-subscribe

Before we get started on this week’s CIW, I’d like to highlight a festive "Chew On This" episode, unwrapping a crucial gift for CIOs and organizations — the wisdom of a proactive incident response strategy.

📅 Save the Date: Wednesday December 13, 2023

🕛 Time: 12:00pm EST

🎙️ Special Guest: Nick Bozovich with Echelon’s Matt Donato and Paul Interval

Discover how a proactive incident response strategy protects sensitive data, ensures compliance, preserves reputation, and reduces costs.

Our experts will guide you through adapting to evolving threats, ensuring business continuity, and maintaining trust amidst the holiday spirit. Don't miss this timely discussion blending security with the season.

Mark your calendar and fortify your organization for the challenges ahead! Register for the third course of Chew on This here: https://lnkd.in/eUdYQA8j

Away we go!

1. ChatGPT's Inadvertent Data Exposure via a "Divergence Attack"

In a recent study, researchers have uncovered a fascinating aspect of large language models like ChatGPT: their ability to inadvertently reveal parts of their training data. Typically, the exact contents of these training datasets are kept confidential, but this new research shows that under certain conditions, models can end up regurgitating specific data they were trained on, a process known as "memorization."

The study, a collaboration between experts from Google DeepMind, the University of Washington, UC Berkeley, and others, aimed to gauge the extent of data memorization in models like ChatGPT. They developed a novel approach to provoke ChatGPT into revealing its training data, termed a "divergence attack." By repeatedly prompting ChatGPT with a specific word, the model eventually diverges from its standard responses, exposing memorized content. This method enabled the team to extract a significant number of verbatim training examples, raising concerns about privacy, especially when personal information is involved.

Notably, the study discovered that as the size of these models increases, so does their tendency to memorize and potentially disclose sensitive information. This revelation calls for a more thorough testing of AI models, not just in their user-facing aspects but in their foundational design and system-wide interactions. It also underscores the need for more robust data handling practices, like training data deduplication, and a deeper understanding of how a model's capacity influences its memorization habits.

This research is pivotal in highlighting a critical aspect of AI development – the balance between powerful language processing capabilities and the safeguarding of user privacy. It opens up new avenues for exploration, urging developers and researchers to devise models that are not only effective but also respectful of privacy concerns. The findings suggest that addressing these issues may require a fundamental shift in how we approach the security analysis of AI systems, treating them akin to traditional software systems with inherent vulnerabilities that need systematic and rigorous scrutiny.

2. Crypto Under the Microscope: The Binance Settlement and its Impact

Binance, the world's largest cryptocurrency exchange, has been brought under the scrutiny of US regulators and law enforcement agencies. This move, following a settlement with the US Justice Department, marks a dramatic shift in the crypto industry, especially for a company known for its lax approach to regulatory compliance.

Binance's journey from a rule-defying startup in 2017 to a global crypto-trading giant with over 100 million users has been tumultuous. The company, initially taking pride in operating beyond the reach of US financial regulations, has now agreed to a settlement involving a record-breaking $4.3 billion fine and a commitment to strict data-sharing with US authorities. This settlement means that Binance will provide access to years of customer transaction data, effectively subjecting the company to continuous and rigorous scrutiny.

This unprecedented level of access to Binance's records by US agencies like the FBI and IRS is a significant change for the crypto exchange, which had previously collected minimal user information. The settlement also requires Binance to retrospectively scrutinize transactions from 2018 to 2022 for potential legal violations, a task overseen by a government-appointed but company-funded monitor. This retrospective analysis is expected to greatly aid in combating illicit activities in the crypto world, potentially transforming Binance into an ally of US law enforcement.

However, this new era of compliance and transparency at Binance is not without its critics. Advocates for financial privacy and digital civil liberties express concerns over the extent of government overreach and the implications for users who chose cryptocurrency for its promise of privacy. This move by the US government is seen as setting a precedent for the regulation of the entire cryptocurrency industry, potentially impacting how users interact with these platforms. Despite the challenges, Binance's compliance could pave the way for more mainstream adoption of cryptocurrencies, albeit in a more regulated and monitored environment.

3. UK Accuses FSB of Sustained Hacking Campaigns

The UK government has accused Russia's Federal Security Service (FSB) of conducting a long-term, systematic cyber-hacking campaign against British politicians and other public figures. This campaign, believed to be ongoing since at least 2015, has primarily involved the theft of private email data, some of which has been publicly disclosed, including material linked to the 2019 UK election. Despite Russia's denials of such activities, the UK's Foreign Secretary David Cameron has strongly condemned these actions as "completely unacceptable."

The UK's response has been firm, with the Russian ambassador summoned and sanctions imposed on two individuals, including a serving FSB officer. The FSB, the successor to the KGB, is believed to have targeted a broad range of individuals, from MPs to academics, through meticulously crafted cyber-attacks. The implicated FSB group, known as Centre 18, has reportedly been active in stealing information for years, using various methods, including creating fake accounts to impersonate trusted contacts.

The accusations against the FSB have sparked serious concerns about the security of the UK's democratic processes. However, Western officials have assessed that these cyber activities have not successfully interfered in democratic processes. The public announcement of these allegations is a strategic move to disrupt the group's operations and increase awareness ahead of significant global elections. This stance by the UK, expected to be echoed by the US, signifies a proactive approach to combating cyber threats and underscores the ongoing challenges of maintaining cybersecurity in the face of sophisticated state-sponsored hacking operations.

Thanks for reading!

About us: Echelon is a full-service cybersecurity consultancy that offers wholistic cybersecurity program building through vCISO or more specific solutions like penetration testing, red teaming, security engineering, cybersecurity compliance, and much more! Learn more about Echelon here: https://echeloncyber.com/about