Post Syndicated from Aaron Wells original https://blog.rapid7.com/2021/08/06/black-hat-recap-2/

Here we are again, back for another day of Rapid7 expert debriefings and analysis for some of the most talked-about Black Hat sessions of this year. So without further delay, let’s take it away!

Detection and Response

Key takeaways

• How do human behaviors — learned or learning — factor into incident response? Depending on the volume of stakeholders, your team may be under varying extremes of action bias. As in, are speedy actions being prioritized on vulnerabilities that don’t present a high risk profile? Is speed even possible if mitigating actions must suddenly be learned? Vendors have caught on, practicing “Security Theater”— peddling solutions to problems that might not present real risks.
• Tangential to the previous topic, a question arises when exploring the weaponization of C2 channels: Due to the unlikelihood of an attack via, say, LDAP attributes when establishing C2, does it make sense to roll out an entirely new detection-and-response plan? Many different conditions must be met for an attacker to gain access in the wild, but teams might already have similar responses in place, on the off chance it happens.
• Zooming out to a topic with broader public appeal, let’s consider how companies use — and abuse — our personal data. An 18-month test run by a professor and a group of students at Virginia Tech revealed how unlikely it is we’ll be able to predict which companies will abuse personal information after someone, say, creates login credentials for a TikTok account and the company launches cookie tracking for that person.

Vulnerability Risk Management

Key takeaways

• Are Microsoft Exchange Servers creating an entirely new attack surface via Client Access Services (CAS)? Exchange architecture is incredibly complex, so it contains multitudes when it comes to vulnerabilities. CAS ties front-end and back-end services together, receiving the front-end request through a variety of protocols, including some extremely geriatric ones like POP3 and IMAP4. These legacy protocols are contributing to expanded attack surfaces.
• Vulnerability Exploitability eXchange (VEX) helps teams rethink security advisories and what it means to be vulnerable. Essentially, it enables software providers to communicate they’re not affected by a vulnerability. Two advantages of VEX are 1) that creation and management of vulnerabilities are automated, and 2) that its results are machine-readable.  
• Open-source software (OSS) is incredible… and incredibly vulnerable. There are so many risks with OSS that a vendor might even put off patching a vulnerability — for whatever business reason — if alerted to it. There’s currently no mechanism to secure so many classes of vulnerabilities in OSS, but maybe there should be. Researchers should work together to create those class-eliminating mechanisms, ultimately reducing the lift when it comes to risk management.

Research and Policy

Key takeaways

• What is Electromagnetic Fault Injection (EMFI)? It’s when hardware attackers use electromagnetism to hack hardware chips. When it comes to something like a car’s modern combustion engine, EMFI can be leveraged to change a vehicle’s performance, slithering past manufacturer-imposed security protocols. Some owners are beginning to “tune” chips with EMFI in order to push the limits of their vehicles.
• There’s cause for concern that AI security products are simply repeating back to us the tables on which they were trained. If this is the case, can someone create more nefarious tables to sway AI security entities away from actual security? Attackers can now train explainable AI models on private data, turning them into the latest tool in their arsenals. Consider your attack surface expanded.
• When companies export their technology beyond their own borders, it isn’t as easy as it sounds in a press release. Whereas policy constantly lagged behind technology, it’s starting to catch up as companies realize the cost of doing business with both digital authoritarians and digital democracies. Is proprietary tech compromised when entering a new country where it must adhere to each and every law imposed on it by local regulators?

Thanks for joining the Rapid7 team at another round of Black Hat debriefings. We hope to see you live and in person in Vegas next year. Until then, stay secure and stay safe!

And if you’re not ready to walk away from the table just yet, revisit our Day 1 takeaways, or sign up now to hear our Research team’s behind-the-scenes insights on DEF CON 2021 at the What Happened in Vegas webinar on Tuesday, August 10.

Post Syndicated from Dwayne A. Johnson original https://blog.rapid7.com/2021/08/05/black-hat-recap-1/

OK, no big deal, we know how this goes. Once again, many of us are attending Black Hat in a virtual capacity as COVID-19 meanders its way out of our lives. The good news is that there’s an actual live component again this year in Las Vegas, and that’s progress. Here’s hoping that next year the pandemic will be more firmly in the rearview and any remaining travel trepidation will be a “2021 thing.”    

So flip the on-switch to some neon lights if you got ‘em, and let’s get into what our Rapid7 experts thought were the biggest takeaways from a busy Day 1 of new tools, techniques, and up-to-the-minute information.

Detection and Response

Key takeaways

• Does it make sense for an organization to “roll its own SIEM”? Yes and no (because of course that’s the answer). For very specific use cases outside of the norm, it might make sense to start the often-herculean, cost-prohibitive task of building that cloud-native SIEM to best serve hyper-specific needs. But is it worth it to miss out on the high-quality, actionable intel a commercial vendor brings to the table?  
• When it comes to distributed malware, attackers are bypassing traditional detection. Return Oriented Programming (ROP — pronounced “rope”) grants attackers a bypass route through initial access points to get onto an endpoint faster and easier. However, the real endgame is to bypass that endpoint agent and hack the network at large.  
• Just how easy is it to hack a hotel? If you were the victim of a hotel hack, you might think a ghost had taken up residence in your room as your IoT-connected bed suddenly moves up and down. However, the proliferation of unprotected networks and IoT devices in modern hotels has created unprecedented opportunities for attackers to gain nefarious access. A back-to-basics approach might be the best way forward for the hospitality industry.

Vulnerability Risk Management

Key takeaways

• Open Platform Communications (OPC) standards are a wondrous thing, allowing products across many industries to interact and exchange data efficiently. But is security a priority? When commercial vendors all along a supply chain start making their own customizations to the common legacy protocol, well, security isn’t so secure anymore.
• Find an active-directory certificate vulnerability? Good luck getting it patched. These configuration-related instances are flaws that larger organizations might be hesitant to acknowledge. Check out this (extremely long, but informative) whitepaper on the subject — and the accompanying blog — from SpecterOps.
• Printer vulnerabilities aren’t paper-thin. Windows Printer Spooler can offer up an attack surface that leads to an instance like the PrintDemon incident. Some of the larger vulnerabilities see attackers and exploit authors leveraging printer path names.  

Research and Policy

Key takeaways

• Let’s talk lasers — specifically, how attackers can use them to exploit vulnerabilities in hardware like bitcoin wallets. One would hope that the key material they’re storing in that wallet is secure. However, with a laser you can “look through” a silicon chip to confuse the CPU and bypass security checks.  
• Wondering how future information wars will be fought? By bots. Advanced bots, that is — those that leverage Generative Pre-Trained Transformer (GPT) language models like GPT-3. With this powerful tool, a small group of people could generate misinformation at scale, quickly spinning up thousands of fake social accounts creating individual posts that sound like actual human language. That’s scary.  
• As far as we know, AI cannot yet be arrested. However, threat actors can still run afoul of digital crime laws like the Computer Fraud and Abuse Act (CFAA) when they employ adversarial machine learning. This “poisoned data” results in systems learning things they shouldn’t. Current federal and state computer-crime laws need to reflect these more sophisticated AI attack methods so that, you know, the machines don’t win.  

We’ll see you right back here tomorrow for Black Hat Day 2 insights and takeaways from the Rapid7 team!