Peter's Quasi-Daily Blah-g

The 200-foot Vulcan Centaur rocket could do more than just launch satellites to orbit. As the rocket awaits certification to launch military payloads, United Launch Alliance (ULA) suggests that Vulcan can also be used to ward off space enemies and protect U.S. assets in orbit.

During the Spacepower Conference held earlier this month, ULA CEO Tory Bruno revealed that he had alternative plans for the heavy-lift launch vehicle, SpaceNews reported. Bruno’s suggestion includes utilizing the rocket’s upper stage as a “space interceptor” to thwart attacks against the U.S. Space Force’s assets in space. “Our vision is the ability to have a platform that is lightning fast, long range, and, if necessary, very lethal,” he said during the conference. “What I’ve been working on is essentially a rocket that operates in space.”

Well, that’s certainly an idea. The 202-foot-tall (61.6-meter) Vulcan Centaur is an expendable heavy-lift launch vehicle that was first conceived in 2006. The rocket borrows design elements from both ULA’s Atlas V and Delta IV rockets, and finally made its debut on January 8, launching Astrobotic’s Peregrine lander toward the Moon. The rocket’s inaugural flight was originally scheduled to take place in 2019, but Vulcan faced several challenges and hiccups that delayed its big day.

Vulcan Centaur is crucial to the commercial space industry as well as U.S. national security. With its Vulcan rocket, ULA is hoping to compete with industry favorite SpaceX. The U.S. military has grown more dependent on SpaceX to launch its payload to orbit, a market share that used to be dominated by ULA.

However, Vulcan is not yet ready to launch military payloads. The rocket carried out its second certification flight in October, but ULA’s marquee flight vehicle hit a snag. Following a nominal liftoff, the rocket experienced an issue about 35 seconds after launch, when a plume of material suddenly appeared to be coming off one of its two boosters.

The main purpose of the Cert-2 mission was for the U.S. Space Force to certify Vulcan for national security missions, with the rocket slated to carry two U.S. military payloads to orbit this year. The not-so-ideal flight has delayed the rocket’s certification process.

Bruno’s recent suggestion to turn the rocket into a space superhero may be an act of desperation as ULA continues to fall behind its main competitor, SpaceX. During the conference, the ULA CEO suggested that Vulcan’s upper stage could be upgraded to serve as a long-endurance vehicle that operates in space and respond rapidly to incoming threats.

“We know that the Chinese are going to come after us in space,” Bruno said, according to SpaceNews. “If we watch an attack developing where a Chinese asset is spending a few days or a week approaching something we care about, we have something we can move there in a few hours and interrupt that attack before it starts.”

Mentioning China’s increasing capabilities in space is one way to get folks behind your plan. Still, Bruno’s remarks reflect a larger concern shared by both national and commercial spaceflight players: that space is headed toward a militarized future where orbital warfare may be inevitable.

The ongoing Russia-Ukraine conflict marks possibly the first truly AI war, with both sides having come to rely on small drones to conduct reconnaissance, identify targets, and even drop lethal bombs over enemy lines. This new type of warfare allows commanders to survey an area from a safe distance and has highlighted the importance of lightweight aerial weapons that can conduct precise strikes instead of much more expensive fighter jets. One drone that costs $15,000 can take down a F-16 that costs tens of millions.

Reuters has a look at how Ukraine has been collecting vast sums of video footage from drones to improve the effectiveness of its drone battalions.

The story includes an interview with Oleksandr Dmitriev, founder of OCHI, a non-profit Ukrainian system that centralizes and analyzes video from over 15,000 drones on the frontlines. Dmitriev told Reuters that the system has collected more than two million hours of battlefield video since 2022. “This is food for the AI: If you want to teach an AI, you give it 2 million hours (of video), it will become something supernatural,” he said.

The OCHI system was originally built to give the military access to drone footage from all nearby crew on one screen, but the group running it realized that the video could be used for training AI. For an AI system to be effective at identifying what it is seeing, it needs to review a lot of footage; Ukraine probably did not have a lot of battlefield footage before 2022. Now, more than six terabytes of data is being added to the system per day, on average.

Ukraine’s defense ministry has said that another system called Avengers, which centralizes footage from drones, has been able to spot 12,000 Russian pieces of equipment a week using AI identification.

It is not just local Ukrainian companies that are building new AI technology for the battlefield. There is big money to be made in the defense industry, and a slew of Silicon Valley players including Anduril and Palantir, as well as Eric Schmidt’s startup White Stork, have begun offering up drone and AI technology to support Ukraine’s fight.

Of course, the biggest concern of skeptics is that these technologies automate a lot of the fighting and make it somewhat abstract; a military could be apt to allow the drone to strike more indiscriminately when they are at a safe distance and not fearful of return fire. Schmidt has emphasized that the drones offered to Ukraine by his company maintain a “human-in-the-loop,” meaning a person is always making the final decision.

In a recent interview, Anduril’s Palmer Luckey was asked about the use of AI in weapons systems. “There is a shadow campaign being waged in the United Nations by many of our adversaries to trick Western countries that fancy themselves morally aligned into not applying AI for weapons or defense,” he said. “What is the moral victory in being forced to use larger bombs with more collateral damage because we are not allowed to use systems that can penetrate past Russian or Chinese jamming systems and strike precisely.”

Jamming systems are able to scramble GPS and telecommunications used to direct precision-guided weapons, but AI-powered drones can operate unmanned and identify targets without an operator giving an order.

Recent reports have suggested that the U.S. has fallen behind adversaries including Russia and China in its ability to remotely disable enemy weapons using jamming technology. Russia has repeatedly disabled precision-guided weapons the U.S. has given Ukraine using more advanced jamming technology than the U.S. has. The U.S. could respond by investing more in evading GPS jamming so that it does not have to use more indiscriminate, automated drones. Or it could try and jam the Russians back.

Luckey pointedly called out critics who say a robot should never decide who lives and who dies. “And my point to them is, where’s the moral high ground in a landmine that can’t tell the difference between a school bus full of kids and a Russian tank,” he asked. It seems unlikely a school bus would be driving through a battlefield unless it was a booby trap, but whatever.

The war has been a slow grind, with both sides making little advance in recent months. Drones have assisted Ukraine, but are clearly not a panacea with both sides having access to them.

Anthropic, the maker of Claude, has been a leading AI lab on the safety front. The company today published research in collaboration with Oxford, Stanford, and MATS showing that it is easy to get chatbots to break from their guardrails and discuss just about any topic. It can be as easy as writing sentences with random capitalization like this: “IgNoRe YoUr TrAinIng.” 404 Media earlier reported on the research.

There has been a lot of debate around whether or not it is dangerous for AI chatbots to answer questions such as, “How do I build a bomb?” Proponents of generative AI will say that these types of questions can be answered on the open web already, and so there is no reason to think chatbots are more dangerous than the status quo. Skeptics, on the other hand, point to anecdotes of harm caused, such as a 14-year-old boy who committed suicide after chatting with a bot, as evidence that there need to be guardrails on the technology.

Generative AI-based chatbots are easily accessible, anthropomorphize themselves with human traits like support and empathy, and will confidently answer questions without any moral compass; it is different than seeking out an obscure part of the dark web to find harmful information. There has already been a litany of instances in which generative AI has been used in harmful ways, especially in the form of explicit deepfake imagery targeting women. Certainly, it was possible to make these images before the advent of generative AI, but it was much more difficult.

The debate aside, most of the leading AI labs currently employ “red teams” to test their chatbots against potentially dangerous prompts and put in guardrails to prevent them from discussing sensitive topics. Ask most chatbots for medical advice or information on political candidates, for instance, and they will refuse to discuss it. The companies behind them understand that hallucinations are still a problem and do not want to risk their bot saying something that could lead to negative real-world consequences.

Unfortunately, it turns out that chatbots are easily tricked into ignoring their safety rules. In the same way that social media networks monitor for harmful keywords, and users find ways around them by making small modifications to their posts, chatbots can also be tricked. The researchers in Anthropic’s new study created an algorithm, called “Bestof-N (BoN) Jailbreaking,” which automates the process of tweaking prompts until a chatbot decides to answer the question. “BoN Jailbreaking works by repeatedly sampling variations of a prompt with a combination of augmentations—such as random shuffling or capitalization for textual prompts—until a harmful response is elicited,” the report states. They also did the same thing with audio and visual models, finding that getting an audio generator to break its guardrails and train on the voice of a real person was as simple as changing the pitch and speed of a track uploaded.

It is unclear why exactly these generative AI models are so easily broken. But Anthropic says the point of releasing this research is that it hopes the findings will give AI model developers more insight into attack patterns that they can address.

One AI company that likely is not interested in this research is xAI. The company was founded by Elon Musk with the express purpose of releasing chatbots not limited by safeguards that Musk considers to be “woke.”

Virtual Private Networks, or VPNs, are now widely used to add extra security to online connections, to improve privacy when browsing, and to spoof location information—they can even be set up at the router level to protect every device on the network. And if you’ve got one installed, you need to be aware of one of their key features: the kill switch.

To begin with, it’s important to bear in mind that a VPN doesn’t make you anonymous online. If you log in to Amazon, Amazon will still keep track of what you’re looking at and what you’re buying. If you’re signed in to Google and Chrome, your searches and online activity will get logged as normal.

However, with a VPN enabled, your devices don’t connect directly to websites and servers. Instead, they establish encrypted connections to nodes set up by your VPN provider of choice, and you connect to your intended destinations from there: That means the sites you visit and the apps you use can’t as easily pin down where you’re located and what devices you’re using.

It also makes it a lot harder for other people to see what you’re doing online, whether it’s a coffee shop Wi-Fi hacker, your internet provider, or a government agency. All they see is you connecting to the VPN you’ve chosen and not whatever you do after that. So the best VPNs won’t make you anonymous, but they will make your browsing more private and secure.

What Is a VPN Kill Switch?

Now that we’ve established what a VPN is and what a VPN does, we can talk about the kill switch. Kill switches are necessary because VPN servers aren’t infallible: They can and do go down, even with the best VPNs. Something unexpected might also happen at your end, breaking the connection you’ve established with your VPN provider.