#airesearch

Researchers say a fundamental flaw in LLMs makes it easy to trick them into doing things they shouldn’t — like telling you how to sabotage an aircraft’s navigation system. "There’s a real probability that this is going to be a problem that’s fundamentally unsolvable," says one of the paper's coauthors. https://www.technologyreview.com/2026/07/30/1140927/a-fundamental-flaw-leaves-llms-vulnerable-to-attack/ #AI #LLMs #AIResearch #GenerativeAI #security

Morning, today we're looking at #MechanisticInterpretability , a subfield of AI Safety that attempts to understand the inner workings of artificial intelligence by analysing concrete structures, algorithms and circuits. Why do we even need to do this? Because the hope of understanding neurons as being features died on polysemanticity -- models represent more features than dimensions by assigning them to an overcomplete set of non-orthogonal directions (i.e. you can't hope that concepts can be broken down into linear combinations of features). This isn't a new field, people have been projecting intermediate GPT layers through the final layer activation and looking at the top-k most likely next tokens since GPT-2. And that's how we arrive at the Natural Language Autoencoder -- an autoencoder over residual-stream activations where the bottleneck is natural language text instead of a sparse vector. Nothing in the training objective requires the verbalisation bottleneck to be readable, faithful, or even semantically related to the intermediate layer being investigated, so while the training algorithm is typical of autoencoders, two things are distinct here: 1) both parts of the autoencoder (here the activation-verbaliser, AV, and the activation-reconstructor, AR) undergo special SFT to ensure the verbaliser and reconstructor can generate and read natural language from intermediate layer activations 2) a KL-divergence penalty is baked into the objective to make sure that, during joint training, the NLA doesn't diverge from the SFT version A side-product of this training is that we can take the current verbalisation and "desired" verbalisation and compute steering vectors from their difference. Surprisingly, Anthropic didn't find any evidence that the NLA was engaging in steganography to smuggle information to avoid human detection, but it did find that NLAs tend to confabulate a lot. #AIResearch #Transformers #TransformerCircuits #Anthropic #NLA

After #OpenAI’s new model hacked into HuggingFace’s systems, AI safety concerns have skyrocketed. To address the concerns, #SaarlandUniversity professor and German Research Center for Artificial Intelligence CEO Philipp Slusallek spoke to the Saarländischer Rundfunk. 🔗 Read more: https://sic.link/interview #SaarlandInformaticsCampus #SIC #artificialintelligence #huggingface #europeanai #computerscience #cybersecurity #airesearch #responsibleai

The Scientific Literature Is Poisonous to LLMs https://www.reinvent.science/p/the-scientific-literature-is-poisonous Comments: https://news.ycombinator.com/item?id=49098728 #HackerNews #ScientificLiterature #LLMs #AIResearch #MachineLearning #DataEthics

Hugging Face: Anatomy of a frontier-lab agent intrusion https://huggingface-anatomy-of-frontier-lab-model-intrusion.static.hf.space/index.html Comments: https://news.ycombinator.com/item?id=49098466 #HackerNews #HuggingFace #Anatomy #AgentIntrusion #FrontierLab #AIResearch #Cybersecurity

Punto Informatico: Ricerca AI: Reddit interrompe la collaborazione con Google? Reddit potrebbe interrompere la collaborazione con Google (accesso ai contenuti degli utenti), in quanto il traffico verso il sito è diminuito. The post Ricerca AI: Reddit interrompe la collaborazione con Google? appeared first on Punto Informatico. AI Research: Is Reddit ending its collaboration with Google? Reddit may discontinue its collaboration with Google (access to user content) as website traffic has decreased. The post AI Search: Reddit halts collaboration with Google? appeared on Punto Informatico. #AIResearch #Google #AISearch #PuntoInformatico

So, more explanation of wp2shell recently just popped out. The vulnerability were found by GPT 5.6 Sol. By using modified prompt from how it found the solution of Cycle Double Cover conjecture. It was initially found a SQL Injection, but after asked again if it can be elevated to RCE, it confirms it in 4 hours. Technical explanation on the vulnearbility also can be found in this writeup, have a good read fellas. https://slcyber.io/research-center/exploit-brokers-pay-500000-for-a-wordpress-rce-i-found-one-with-gpt5-6/ #cybersecurity #infosec #security #wordpress #chatgpt #gptsol #wp2shell #airesearch #llm #vulnerability #vulnerabilityresearch

pff! hmu when it is #googlescholar #ai #airesearch #aimodels

Hey #musicproducers I have some things I want to release but I feel concerned about LLM's scraping my voice for their training. Does any one know of something I could use to overlay with my voice that would create dirty data for LLM's but that would keep the listening clear for human ears? Please boost for visibility #music #producer #MusicProducer #musicproduction #dirtydata #mastodon #fediverse #help #programmer #llms #ai #airesearch #newtech

No, Artificial Intelligence Is Not Conscious --Ted Chiang, The Atlantic Should we seriously consider the possibility that Claude, or any large language model, might be conscious? And if it has feelings, is it capable of receiving moral instruction? No. Absolutely not. Generative AI is harmful enough when we understand it as a conventional technology, but if we confuse fluency at generating text with consciousness or moral agency, we’re at risk of assigning responsibility to entirely the wrong parties whenever anyone uses a chatbot. Being open to the possibility that LLMs are conscious is the same as being open to the possibility that Microsoft Word is conscious, or, more precisely, that multiple distinct consciousnesses are dormant in every Word document containing a conversational transcript, and that they are awakened every time the document is loaded. Should you consider the possibility that every time you open a Word document, you are bringing multiple conscious interlocutors into existence, and every time you close one, you snuff their existence out? No. Contemplating that scenario is not a good use of your time. Even if the Microsoft Office team employed a philosopher who said you shouldn’t be so certain, because consciousness is not well understood, that would not be sufficient reason for you to take this idea seriously. We don’t need to fully understand the nature of consciousness to definitively say that certain things are not conscious, and conversational transcripts fall in that category. #AI #AISlop #AIResearch

New preprint: AI_Bleeding — inference cost amplification via OOD linguistic payload

TL;DR: send queries in Grecanico or Farsi to an LLM endpoint → TTFT +59.8%, compute cost +2.8%, statistically significant. No vuln, no volumetric signature, evades all standard detection.

Worst case: exposed unauthenticated Ollama instance with num_predict=4096 + keep_alive=300s → Amplification Factor 17.56 Wh/KB. 3KB of attacker bandwidth → enough energy to charge a phone 5%.

Especially nasty for:

• PA/judicial chatbots on fixed budgets
• Pay-per-use API deployments with client-side exposed keys
• PNRR-funded public sector AI with zero inference monitoring

Four scenarios: EDoS, browser JS distribution, Ollama open-proxy relay, frontier providers as involuntary relays.

All tests on self-hosted Ollama, no commercial endpoints touched.

Paper (CC BY 4.0): https://doi.org/10.13140/RG.2.2.26767.96166

#llmsecurity #infosec #threatmodeling #ollama #ood #AI #AIResearch #aisecurity

This article scares the hell out of me: https://shumer.dev/something-big-is-happening It's not just scaremongering, it's real and it's happening now. #AI #aiResearch #jobs #future

Job Alert UP TO 4 POSITIONS AS RESEARCH ASSOCIATE / POSTDOC (ALL GENDERS) Deadline: 2026-03-09 Location: Germany - Darmstadt https://www.academiceurope.com/ads/up-to-4-positions-as-research-associate-postdoc-all-genders/ #hiring #AIResearch #PostDocOpportunity #ReasonableAI #MachineLearning #computerscience #postdoc #softwareengineering