Reposted by Running With SpoonsVictoria Strauss @victoriastrauss.com · 6hCanada is refusing to issue ISBNs to books primarily generated with AI. This illustrates both the urgency of the problem of the growing flood of slop books, and also the toothlessness of addressing it on the honor system without there also being consequences for lying 151861510
Running With Spoons @runswithspoons.bsky.social · 6hCreating a clean room spec by having a chatbot randomly click around an app until it completely understands the functionality Or maybe Creating a clean room spec by having a chatbot read the online manual and randomly guess at all the details not in the manual 010
Running With Spoons @runswithspoons.bsky.social · 06/10/2026I worked as a developer in the 00s but moved to different roles when I saw the push to offshore dev work. This LLM push feels similar. My experience is that mgmt aggressively outsources for cultural reasons. You have to keep tech people happy to retain them. You have to give them a voice. 010
Running With Spoons @runswithspoons.bsky.social · 03/10/2026Exclusive: A Ouija board in Northern Cleveland grants the power to speak with the dead. To get to the bottom of this perplexing behavior, Science used the Ouija board to interview a ghost. 010
Running With Spoons @runswithspoons.bsky.social · 02/10/2026dl.acm.org/doi/abs/10.1... The newer approaches to this rely especially on beam forming, since there are multiple antennas and the hardware is already designed to monitor signal strength, angle, etc Just another reason not to fill your home with wifi devicesdl.acm.orgEnabling Ubiquitous WiFi Sensing with Beamforming Reports | Proceedings of the ACM SIGCOMM 2023 Conference 120
Reposted by Running With SpoonsHypervisible @hypervisible.blacksky.app · 02/10/2026“AI companies are fixated on adding features right now…But as always, the more features, the broader the attack surface. So all of these companies need to be fully focused on security, and from what I can see, it still often seems like an afterthought.”wired.comA Flaw in ChatGPT’s Mac App Could Have Let Hackers Grab Sensitive DataWhile the focus has been on AI agents’ hacking capabilities, a recently patched vulnerability in a ChatGPT app shows that AI software is itself an inviting—and vulnerable—target. 07031
Running With Spoons @runswithspoons.bsky.social · 01/10/2026Maybe this is zuckerberg dumping money to try to take over the market? 000
Running With Spoons @runswithspoons.bsky.social · 30/09/2026When I quit Amazon a year and a half ago I worried I'd have trouble finding the things I need. What did I find? My impulse purchases dropped to near zero and I have a *lot* less cardboard to recycle 0190
Reposted by Running With Spoonsevacide @evacide.bsky.social · 28/09/2026Meta Muse appears to read your Apple messages and upload them to the cloud even if you explicitly tell it not to: appleinsider.com/articles/26/...appleinsider.com 10220331029
Running With Spoons @runswithspoons.bsky.social · 28/09/2026"Guys no one doubts that OpenAI and Anthropic are responsible if their models go rogue." ☝️Legally speaking this is certainly not true. Also, why are you running cover for US Big Tech billionaires? 010
Reposted by Running With SpoonsLex Roman @lexroman.com · 25/09/2026Watch what you post because Threads just might pretend you're the guy who masturbated on the plane. What just happened to Peter Shankman is a warning for all of us using Meta products. 321701507
Reposted by Running With Spoonsjonny (leachate composite form) @jo.nny.rip · 24/09/2026The #meta #muse VM is trivially dumped with casual chat prompt injection, and all the "get started" ideas that won over all the hype bloggers, including the new york times, are fucking hardcoded neuromatch.social/@jonny/11732... 475
Reposted by Running With Spoonsevacide @evacide.bsky.social · 26/09/2026Listen, I just think it's weird that the same people who are saying they're worried that AI is going to wipe out humanity are also eager for military contracts that will give AI control of weapons. 431433341
Running With Spoons @runswithspoons.bsky.social · 24/09/2026Thank you, I'll try this suggestion with our team developing the chat interface. They mentioned having a smaller LLM filter chat output but were concerned about the latency, which they said was 2-3 seconds in their testing. 000
Running With Spoons @runswithspoons.bsky.social · 24/09/2026They found other vulns that might enable content injection or request forgery attacks. Possibly another jailbreak to put the chatbot in admin mode. The solutions we have are to add instructions to the system prompt telling it not to do this. Or to ban keywords from user input or chat output. 130
Running With Spoons @runswithspoons.bsky.social · 24/09/2026Using the jailbreak, our pen tester was able to have the chatbot answer embarrassing questions about the company (e.g. publicly reported data breaches), provide technical knowledge about how to run ransomware attacks, summarize US political stories, list its internal tools, etc. 130
Running With Spoons @runswithspoons.bsky.social · 24/09/2026Yes good question. In this case, it's a read-only chatbot with a very limited set of actions. Our pen tester was able to use a context attack to jailbreak the chatbot and have it answer questions using public information. The risk (for now) is mostly reputational. 130
Running With Spoons @runswithspoons.bsky.social · 24/09/2026A way to explain what happened is that OpenAI and Anthropic are developing automated weapons but neglected to test them safely. Imagine if someone was developing armed drones in a vacant lot in your city and one of them flew away and blew up a local store. 000
Running With Spoons @runswithspoons.bsky.social · 24/09/2026My employer is currently rolling out a customer-facing chatbot and I'm one of the infosec folks trying to secure it. I'm only seeing two strategies: add instructions to the system prompt or put a filtering tool in front of it. Feels like a lost cause. Get ready for a decade of these problems. 252
Reposted by Running With SpoonsAP Stylebook @apstylebook.com · 23/09/2026Artificial intelligence systems do not think, feel, want or understand. Avoid language that gives them human characteristics.apnews.comhttps://apnews.com/article/openai-safety-ai-framework-089e75b95bc935af092da7b79d92706d 5542401630
Running With Spoons @runswithspoons.bsky.social · 24/09/2026Maybe LLM chatbots intended for everyday office work should not be trained on all the hacking materials and reddit posts OpenAI manages to steal? Alignment is a fantasy when you train your LLMs on deception. (LLMs are not the right technology for any of this work) 000
Running With Spoons @runswithspoons.bsky.social · 23/09/2026Thank you sir can we please ban all the other fairy tales too? Claude told me about people who turn to wolves in the full moon and this sounds very serious 000
Running With Spoons @runswithspoons.bsky.social · 23/09/2026Those floppy drives worked some of the time and then the games were pretty good 010
Running With Spoons @runswithspoons.bsky.social · 22/09/2026futurism.com/artificial-i... Working in corporate USA used to be a game of "don't step on the land mines" but now we have to dodge slop grenades toofuturism.com"Slop Grenades": Shopify CEO Who Pushed Staff to Use AI Is Now Horrified by What He's WroughtShopify CEO Tobias Lütke is getting a first-hand look at the AI-workplace revolution, and it isn't a pretty sight. 000
Running With Spoons @runswithspoons.bsky.social · 22/09/2026If you could build strong AI by "scaling up" the vacuum cleaner of stolen training data, all this human training wouldn't be necessary. The LLM vendors are out of ideas already 000
Running With Spoons @runswithspoons.bsky.social · 22/09/2026OpenAI hired contractors to review LLM output but those contractors were using LLMs so OpenAI had to hire more contractors to review the output of the first group of contractors to make them stop using LLMs. But those contractors are probably using LLMs too 110
Running With Spoons @runswithspoons.bsky.social · 22/09/2026This is where I make an excuse to leave the conversation and go talk to someone else at the bar 000
Running With Spoons @runswithspoons.bsky.social · 21/09/2026If that's what you took away from the post then you chose to miss the point 110
Reposted by Running With SpoonsEd Newton-Rex @ednewtonrex.bsky.social · 19/09/2026“We trained GPT-3 on pirated stuff! No sharing that!” Astonishing admissions in these newly released documents in Authors Guild v OpenAI. www.courtlistener.com/docket/67810... 23015
Reposted by Running With SpoonsRima I Anabtawi @rimaanabtawi.bsky.social · 18/09/2026The US military was prepared to board a Chinese vessel in the ME earlier this spring -- during war with Iran -- after intel report concluded it was transporting components of a nuclear weapons program. the report had been generated w/ help of AI & chatbot inaccurately www.cnn.com/2026/09/18/p...cnn.comExclusive: US military had close call after using AI for false intelligence report, sources say | CNN PoliticsThe episode shows the risks of using this new, relatively poorly understood technology in the middle of the Iran war 0124
Running With Spoons @runswithspoons.bsky.social · 18/09/2026techcrunch.com/2026/09/18/r... LLM vendors position their products as cybersecurity supertools. If that's true, why is their security so bad?techcrunch.comResearchers used Anthropic's Claude to hack into OpenAI | TechCrunchSecurity researchers used Anthropic’s Claude to exploit vulnerabilities in OpenAI’s systems, taking over employee accounts and gaining access to an internal code repository before reporting the flaws. 000
Running With Spoons @runswithspoons.bsky.social · 18/09/2026Among other problems, they won't catch design or architectural flaws until they're testing a finished codebase. Assuming they catch those flaws at all 010
Reposted by Running With SpoonsJed Brown @jedbrown.org · 12/09/2026A good example of the misleading marketing by CoCounsel not being accepted by the court, which issued sanctions and a fine. websitedc.s3.amazonaws.com/documents/Hi... 253
Running With Spoons @runswithspoons.bsky.social · 17/09/2026Oh come on. It *is* true, and you need to admit to yourself you have no idea how this technology works 000
Running With Spoons @runswithspoons.bsky.social · 16/09/2026He misrepresents stats to justify vibes. "Yes I said 85% chance of A, then B happened. But I wasn't predicting the outcome, I was predicting likelihood" 🤔 Not really falsifiable unless you run 1000 experiments Unsurprisingly his fans are as bad as elon's. Recommend avoiding them 180
Running With Spoons @runswithspoons.bsky.social · 16/09/2026"They've always been able to put guardrails in." Eventually people will realize this isn't quite true. That won't help the bubble 010
Running With Spoons @runswithspoons.bsky.social · 15/09/2026Fwiw, deception and lies are almost certainly part of the offensive security training data. So the alignment efforts are working against the training data. 010
Running With Spoons @runswithspoons.bsky.social · 13/09/2026Would you trust your cat with a machine gun 130
Running With Spoons @runswithspoons.bsky.social · 13/09/2026What a gift to attackers, who can afford to move fast and break things. When cyber defenders "break things" we might take down the org we're defending. "Claude, hack this website" is safe for attackers. "Claude, isolate any system with signs of a breach" can get a defender fired 100
Running With Spoons @runswithspoons.bsky.social · 13/09/2026The solution LLM vendors offer is to outsource all that cyberdefense work to their chatbots. But if their chatbots are so good at defense, why do LLM vendors keep having security problems? It's hard to overstate how incompetent these companies look at securing their own networks. 100
Running With Spoons @runswithspoons.bsky.social · 13/09/2026Small organizations can't afford to hire the professional monitoring and incident response teams they need to actively defend their networks. They can't afford millions of dollars each year for software tools and teams of people to find and fix all the problems. 110
Running With Spoons @runswithspoons.bsky.social · 13/09/2026From what I've seen, they've mostly *raised* costs for defenders. We have to move faster with patching old systems, we have to build better monitoring. We have to spend a fortune on new tools to protect LLMs from attacks. 110
Running With Spoons @runswithspoons.bsky.social · 13/09/2026What Anthropic and OpenAI have done is to make nation-state level attacks more accessible to people without a team of PhDs in Computer Science and a multimillion dollar budget. 100
Running With Spoons @runswithspoons.bsky.social · 13/09/2026We had priced out individual attackers and many small teams from carrying out substantial attacks because they didn't have the resources. Gone were the days of hackers "in their mom's basement". It took a team with training and funding. Cybercriminal orgs are run like professional businesses. 100
Running With Spoons @runswithspoons.bsky.social · 13/09/2026When we share threat intelligence, the purpose is to increase attacker costs. We force them to rent new servers and buy new domains by blocking known IPs and domains. We force them to build new tools and develop new playbooks because we countered the old ones. These things cost money and time 100
Running With Spoons @runswithspoons.bsky.social · 13/09/2026Anthropic admits its LLMs have reduced the cost to run professional cyber attacks. People outside the industry might not realize how destabilizing this will be. A key strategy in defense is to shift cost onto your attackers. It has been effective. www.anthropic.com/threat-intel... 120
![At the Order to Show Cause hearing, Attorney Webb averred that his namesake law firm
had conducted training about the problems with misuse of AI prior to the firm’s use of Westlaw’s
CoCounsel AI tool. [Dkt. 197 at 4]. He averred that his firm’s staff have daily ongoing didactic
trainings and again referenced sending staff to the ClioCon conference. Id. at 4–5. Attorney Webb
acknowledged reading the Thompson Reuters disclaimer about the accuracy of the CoCounsel AI
tool, but emphasized that the sales representative stated there was “no way” the program could
produce hallucinations. Id. at 6–7.
Case 3:23-cv-06558-PHK Document 230 Filed 04/28/26 Page 11 of 17](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:rbjag77ol7yqto2ph5gxttf2/bafkreieghgndpkafdqwwno47acw6apsipox6mz7jes2n33seev2plpbeb4)
![jq query of all the hardcoded ideas titles
jq '.["ideas"][].title' ./home/hatch/assets/ideas/new-user/catalog.json
"I can find and cancel those random subscriptions you forgot about."
"Get back money that's already yours. I'll check every state's database."
"I can get you money back when a price drops on an item you bought."
"I'll pick the best rewards card to use and switch on the offers."
"Set a savings goal. I'll flag what to cut when you go off track."
"I can turn your inbox into a complete plan for the day."
"Let me triage your inbox down to only what matters."
"I'll build you a fresh personality quiz every day."
"Snap a pic of what's broken. I'll find the part you need and order it for you."
"New home project? I'll gather the quotes and find the best vendors."
"Flag a pothole or busted streetlight. I'll try to get it fixed."
"Moving? I'll get quotes, book the movers, and file your address change."
"Tell me the item. I'll find the best-reviewed version at the best price."
"Let me hunt FB Marketplace for you. I'll haggle the price and arrange the pickup."
"Tell me what you're buying. I'll show you a side-by-side view of every option."
"I'll book the doctor follow-ups you keep putting off."
"Name a habit you want to break. I'll nudge you before your trigger."
"Show me the gear you've got. I'll build your whole workout around it."
"Snap a pic of your meal. I'll track every calorie."
"Gather the family. I'll host a live trivia night."
"Losing touch with someone? I'll find a time and draft the invite."
"Tell me about your pet. I'll keep their shots and food orders on schedule."
"I can plan your trip and track it live if anything changes."
"I'll track your reservation and make sure the crib, late check-out, and food allergy request are all set."
"I track your flights and warn you the moment anything changes."
"Fare drops? I'll book the seat as soon as it hits your budget."
"Tell me the vibe. I'll book the perfect spot."
[length restrictions, should be OCR-able]](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:t5atl3zfywjjcsutabwimnk4/bafkreiaitoncihynoq7xzlocjqpdutox7qadv6megqqyzfgxz4toz6tl6a)
