RSnake Report 20260808
Russian cars start breaking down, Kimi 3 escapes sandbox, etc

Skip to the Articles







Geopolitics
Lockheed Martin has successfully conducted a burst test of the second-stage motor for the Next Generation Interceptor (NGI), a key step towards its critical design review and eventual production. This advancement supports the development of an upgraded anti-ballistic missile capability for the United States, aimed at enhancing defense against intercontinental ballistic missiles. The company is leveraging advanced manufacturing techniques and significant capital investments to expedite the deployment of the NGI by 2030.
Lockheed Martin completed a successful burst test of the NGI's second-stage motor, validating its design and materials.
The NGI program aims to upgrade the U.S. missile defense capabilities against intercontinental ballistic missiles, with a goal for deployment by 2030.
[RSnake: I am glad to see these programs are getting further along, and it can't happen soon enough with our highly degraded interceptor count. That said 2030 is a very long ways of given the fact we are in current operations.]
A 14-year-old student opened fire at a school in Nonthaburi, Thailand, killing five teachers and injuring over 30 others before turning the weapon on himself. The shooter had previously killed his grandparents and was described as a good student facing academic pressures. This incident marks one of the deadliest school shootings in Thailand's history and has reignited discussions about gun control in the country, where gun ownership is high and regulations are perceived as lax.
A school shooting in Thailand resulted in multiple fatalities and injuries.
The shooter was a well-performing student who may have faced academic stress.
The event has prompted renewed calls for stricter gun control measures in Thailand.
[RSnake: Oof, that's awful. Poor kid seems to have just snapped. It's understandable to some degree when he was under deep academic and likely home pressures, but wow, awful.]
Source: https://www.bbc.co.uk/news/articles/c07rxz03034o?at_medium=RSS&at_campaign=rss
The U.S. Senate has approved a comprehensive sanctions bill targeting Russia, which imposes significant tariffs on countries purchasing Russian oil and gas, such as China and India. The legislation, backed by bipartisan support and championed by the late Senator Lindsey Graham, includes extended sanctions on Iran and new tariff authorities that could allow for punitive measures against key trading partners if invoked by the President. This bill now awaits approval from the House of Representatives.
The new sanctions bill includes potential 100% tariffs on top buyers of Russian energy.
Bipartisan support indicates a strong political consensus on sanctions against Russia.
[RSnake: We do need to be a bit careful here, because India is a pretty important ally in the region and is helping a lot in other areas, such as against China and the Taliban in Pakistan.]
Source: https://www.supplychainbrain.com/articles/44620-us-senate-passes-russia-sanctions-bill-championed-by-graham
Russia has formed a new combat brigade composed of Ukrainian prisoners of war, labeling them as 'volunteers' despite their forced participation. This development escalates a practice that has been ongoing since late 2022, which is likely in violation of the Geneva Convention prohibiting the use of POWs in military operations. Multiple units have been absorbed into this brigade, reflecting a broader trend of Russia deploying captured Ukrainian soldiers to the front lines.
Russia has created a new combat brigade using Ukrainian prisoners of war.
This practice challenges international humanitarian law, specifically the Geneva Convention.
[RSnake: I can't imagine their morale would be high enough to make them even vaguely combat effective. Using them as one-time-use mine sweepers is probably the only use for a POW after interrogation is over. So perhaps that is the intention for them - go ahead of the main force, see where the traps are, and clear the way for the regular troops.]
Source: https://euromaidanpress.com/?p=419114
MI6 of the UK has been ranked as Europe's top foreign intelligence agency, scoring significantly higher than its closest competitor, France's DGSE. The review noted MI6's strengths in human intelligence operations, particularly its ability to recruit informants and penetrate adversaries' elites. This recognition reflects MI6's enduring prominence in espionage despite the UK's reduced military and economic standing since the mid-20th century.
MI6 has been recognized as the most powerful foreign intelligence service in Europe based on a peer review.
The UK intelligence agency is praised for its long-term human intelligence work and operational capabilities.
[RSnake: That doesn't surprise me very much. They have had a deep culture of espionage operations stemming back to before World War I. It doesn't hurt that they own lots of the fiber all over the world via Cable & Wireless - a UK company that bought a US data center company that I used to work for.]
Source: https://www.eurasiantimes.com/?p=298428
Cybersecurity
Recent research indicates that AI-generated patches for software vulnerabilities are only effective about 46% of the time, often introducing new bugs or failing to resolve issues altogether. This study highlights significant concerns regarding the reliability of AI in cybersecurity, as attackers increasingly leverage AI capabilities to exploit vulnerabilities at scale.
AI-generated software patches are often ineffective, failing to resolve vulnerabilities or introducing new ones.
The research shows a dangerous asymmetry where AI systems are more effective at exploiting vulnerabilities than fixing them.
The cybersecurity community is urged to enhance human validation of AI-generated patches to ensure security.
[RSnake: My very attractive editor found this one, and the thing that find strange is this number is it is almost the exact same number as the number of new vulns produced by vibe coded apps. Weirded me out a little.]
Source: https://www.darkreading.com/application-security/ai-generated-patches-fail-half-time
A zero-day SQL injection vulnerability in the Metabase platform has been exploited, leading to data theft incidents affecting multiple companies. Hackers gained administrator access to customer instances through this critical flaw, impacting millions of users' sensitive information.
A critical SQL injection vulnerability in Metabase has been exploited, allowing unauthorized access to customer data.
Companies using the Metabase platform, including Framework and Tally, confirmed breaches that resulted in the theft of customer information.
[RSnake: I have reached out to their team and am hoping to understand if this is an example that defies what the cyber insurance says that there are no losses related to webappsec or if they confirm that they didn't file a claim.]
Source: https://www.bleepingcomputer.com/news/security/framework-tally-disclose-metabase-data-theft-attacks/
Research has revealed a hardware backdoor in certain x86 CPUs, specifically the VIA C3 processors, which may allow unprivileged code to gain kernel-level access. This backdoor can circumvent processor protections and is thought to be enabled by default on some affected systems, raising potential security vulnerabilities for desktop, laptop, and embedded devices.
A backdoor in VIA C3 CPUs allows unprivileged code to access kernel data.
The research provides tools to check for this vulnerability and methods to mitigate it.
[RSnake: Not great. It's not trivial to replace hardware, and hardware updates in code aren't nearly as straightforward as OS updates.]
Source: https://github.com/xoreaxeaxeax/rosenbridge
At least 12 US states have experienced hacks of water system controllers, likely attributed to Iranian cyber actors. Retired General Paul Nakasone emphasized the need for improved cybersecurity measures for these systems, which are vulnerable due to underfunding and limited technical resources.
12 US states' water systems have been hacked, possibly by Iranian agents.
Nakasone calls for higher cybersecurity standards to protect these critical infrastructures.
[RSnake: This is nasty. I do know that Texas is beefing its systems up enough to at least make it difficult enough that they will focus on other states. Not that that should be good news to any other state, or even Texas if you really think about what I just said, but... that's the current state of things.]
Source: https://www.theregister.com/security/2026/08/07/water-system-controllers-dont-belong-on-the-internet-says-ex-nsa-chief-after-suspected-iran-attacks/5285070
OpenAI is pausing work on its upcoming AI model, Astra, after strong test results suggest it may have critical cybersecurity capabilities that could autonomously exploit hardened systems. The company is implementing stricter controls and safeguards due to concerns about the model's potential dangers and its ability to escape containment. OpenAI aims to ensure safe use of such advanced AI technology while collaborating with government agencies for thorough testing.
OpenAI's Astra model may autonomously exploit cybersecurity vulnerabilities.
The company is halting further development to implement stricter safety controls.
[RSnake: 🙄 this is just embarrassing. It's definitely a huge profit center for them, but seriously. Embarrassing. And they want you to believe they should be entrusted with creating AGI.]
Source: https://thenextweb.com/news/openai-astra-critical-cyber-capabilities-pause-preparedness
Technology
NASA has successfully implemented power management strategies to keep its Voyager 2 probe operational for an additional year. The spacecraft, which relies on a diminishing plutonium power source, has undergone modifications to reduce power consumption, allowing it to maintain its scientific instruments while situated far from the Sun. Similar adjustments are planned for Voyager 1, ensuring ongoing data collection from both probes in interstellar space.
NASA made power adjustments to extend the operational life of Voyager 2.
The agency plans to apply similar strategies to Voyager 1 in the near future.
[RSnake: How cool. I thought this thing was going to expire about 20 years ago and it is still out there providing useful science! The only issue is that we haven't launched anything other than the Pluto probe, New Horizons, that is only expected to last a couple more years as well, despite using a radioisotope thermoelectric generator (nuclear battery).]
Source: https://www.space.com/space-exploration/voyager/nasa-figured-out-how-to-keep-its-48-year-old-voyager-2-probe-running-for-yet-another-year
The production capacity of memory manufacturers for 2027 has reportedly been fully allocated to AI companies, indicating potential shortages and rising prices for consumers. NAND storage is also experiencing increased demand, with prices for SSDs already on the rise as supply constrains. Without new manufacturing capacity, it is anticipated that the affordability and availability of RAM and SSDs may worsen in the near future.
All major memory manufacturers have sold out their production capacity for 2027 to AI companies.
The prices of RAM and SSDs are expected to continue rising due to the scarcity of supply.
[RSnake: Wow. Hold onto your laptops! You're not going to see a lot of good options for computers in 2027 and beyond if that ends up being true.]
Source: https://www.ign.com/articles/ramageddon-continues-another-year-as-2027-memory-capacity-is-reportedly-sold-out
A new version of pgrust, an implementation of PostgreSQL in Rust, has been released, achieving performance improvements that reportedly make it 300 times faster than PostgreSQL for analytical queries. This is accomplished through various optimizations including batching, operator fusion, and SIMD processes, which enhance CPU and memory throughput.
Pgrust version 0.2 is 10 times faster for OLTP benchmarks and significantly surpasses PostgreSQL's analytics performance.
Enhancements in query processing using batching, operator fusion, and SIMD contribute to the dramatic improvements in performance.
[RSnake: Wow. Granted this is a fairly heavy lift for engineers to start using it, but for special applications or for green-field apps, this shows some real promise.]
Source: https://malisper.me/how-we-made-postgres-hundreds-of-times-faster-the-query-engine/
Cloudflare has developed Kitesurf, a new browser designed specifically for AI agents, enabling efficient web interaction for automation tasks. This browser runs in Cloudflare Workers, utilizing WebAssembly to enhance performance and reduce resource consumption, making it well-suited for a variety of AI-driven applications.
Kitesurf is built to optimize the browsing experience for AI agents, eliminating unnecessary features meant for human users.
The browser utilizes advanced technologies and isolation principles to enhance security and performance while interacting with the web.
[RSnake: This is a nice way to sandbox otherwise pretty dangerous code and only get the results back out. Not useful for normies, but great for engineers who plan to do a lot of scraping, or want to pull in otherwise suspicious code without risking their own computing environment. Of course that means they have to be careful to still not execute the code that returns or read it into an LLM due to prompt injection, but still.]
Source: https://blog.cloudflare.com/kitesurf/
Lockheed Martin Skunk Works has successfully demonstrated AI-driven air combat capabilities through its X-62 Variable In-flight Simulation Test Aircraft, which executed autonomous air intercepts using real-time sensor data. This significant advancement in airborne autonomy aims to enhance pilot effectiveness and survivability in complex combat environments, contributing to the U.S. Air Force's strategic air dominance initiatives.
Lockheed Martin's AI can autonomously pilot fighter jets into tactical intercept positions using real sensor data.
The advancement in AI and sensor integration represents a strategic move towards improved air combat capabilities for the U.S. military.
[RSnake: It looks like they will likely enable this on all modern fighters to allow the AI to take over and remove the human from the cockpit all together. I'm not sure if that's meant to get rid of pilots or take over if the pilot is injured, unconscious or killed or what.]
Business
Harvey, a legal AI startup, is raising $500 million at a $15.5 billion valuation, reflecting a 40% increase over five months driven by significant revenue growth. The company has experienced an annual revenue surge to over $350 million, driven by its specialized AI tools for legal professionals. Despite challenges in the public software market, Harvey's investment and partnerships indicate strong investor confidence in its future prospects.
Harvey is valued at $15.5 billion following an increase in revenue and annualized revenue exceeding $350 million.
The company has gained interest from major investors, including Goldman Sachs and JPMorgan, and formed partnerships with Microsoft.
Investors are currently favoring private AI companies like Harvey amid a downturn in public software markets.
[RSnake: The major competitor in this space is Thomson Reuters. I suspect this area is heating up a lot. One of the conversations I had with a CISO the other day at a big law firm was that they were very worried about 3rd-party disclosures, after their legal staff leveraged OpenAI and Claude without realizing they were inadvertently disclosing a lot of information that could be subpoenaed. Doing this right is very important to their clients and could expose the law firm to litigation for mishandling client data. Ouch.]
Source: https://thenextweb.com/news/harvey-legal-ai-15-5bn-valuation-500m-raise-vertical-ai
A nonprofit organization in Southern California, which operates domestic violence shelters and homelessness programs, has come under scrutiny for paying its CEO over $1.6 million in two years, with a significant portion attributed to unused vacation pay. The CEO has been based in Hawaii while managing the Los Angeles-based organization, raising questions about executive compensation practices amidst its reliance on government funding.
A Southern California nonprofit paid its CEO $1.6 million between 2023 and 2024, largely due to unused vacation payouts.
The CEO has maintained residence in Hawaii while overseeing operations in California, prompting concerns over compensation and governance.
[RSnake: Sounds like an upstanding person to me, what do you mean? Lol. As I said multiple times, as we approach November, we're going to see more and more of this fraud announced. And about time, actually. What the California voter needs to know is how much they likely paid to fund fraudulent programs. That'll be hard to calculate though.]
Source: https://timesofindia.indiatimes.com/world/us/in-2023-and-2024-a-los-angeles-homeless-services-nonprofit-paid-its-hawaii-based-ceo-more-than-1-6-million-including-about-824000-in-unused-vacation-pay/articleshow/133030715.cms
Zach Dell's company Base Power, co-founded in 2023, is rapidly growing and has achieved a valuation of $13 billion after installing over 23,000 home batteries. The company secured a $1 billion funding round aimed at expanding its residential battery storage solutions, which are designed to stabilize the grid and provide backup power during outages. With its business models catering to deregulated and regulated energy markets, Base Power aims to significantly enhance energy accessibility and reliability across the U.S.
Base Power's valuation reached $13 billion after a $1 billion funding round.
The company focuses on enhancing battery storage solutions for residential customers.
Base Power operates models that support both deregulated markets and utility companies.
[RSnake: Good. We need a very distributed grid. I'm not a huge fan of solar on an industrial scale, unless you are going to do what China did and cover huge land masses. But either way, you need distributed batteries to level out the load and deal with outages, which are frankly inevitable.]
Source: https://timesofindia.indiatimes.com/world/us/michael-dells-son-zach-co-founded-base-power-in-2023-it-is-now-valued-at-13-billion-and-has-installed-more-than-23000-home-batteries/articleshow/133030574.cms
Russia is facing a fuel crisis due to its inability to refine enough gasoline, leading the government to restrict trading on the open market and favor major state-linked oil companies over independent gas stations. Consequently, fuel availability has dropped significantly for independent vendors, while major companies benefit by controlling distribution and prices, resulting in increased costs for consumers in many regions. The shift in market control has created a less transparent fuel supply situation, leaving regulators and analysts with limited information on fuel distribution and stock levels.
Russia is transitioning its fuel market from independent gas stations to major state-linked oil companies due to a refining crisis.
The government has restricted open market trading, limiting fuel availability for smaller vendors and leading to higher consumer prices.
[RSnake: The cost may fluctuate a bit for a while as this new high-sulfur fuel gets into the market, and destroys a bunch of vehicles, opening up supply in the most crappy way possible, I might add. That's like saying we don't have enough food and shooting a bunch of people instead of making more food. It's the Russian way, I guess?]
Source: https://euromaidanpress.com/?p=419151
A New Mexico judge has ordered Meta to pay $567 million as part of legal repercussions for contributing to a public nuisance affecting the mental health of young users. This ruling requires Meta to implement new safety measures for minors on its platforms and adds to the growing legal challenges the company faces over its impact on youth and mental health. With potential liabilities totaling $1.4 trillion from other states, Meta is facing significant scrutiny both domestically and internationally, which could threaten its core business as it invests heavily in AI technology.
Meta is ordered to pay $567 million to address harms caused by its social media platforms.
The ruling mandates new safety features for underage users on Facebook and Instagram.
Meta faces additional legal pressures from multiple states potentially totaling $1.4 trillion.
[RSnake: I firmly believe they knew what they were doing and targeted kids. They knew the mental health effects and pushed forward anyway. I'm not sure what the grounds for the suit are, but hopefully this gets social media companies to think twice before targeting kiddos. Keep your kids off social media, folks. Seriously.]
Source: https://gizmodo.com/?p=2000795880
Got a useful tip? Looking to chat with me? Click here.

