“Breakthroughs in large language models (#LLMs) have made conversations with #chatbots feel more natural and human-like. Applications such as Replika and Character.ai, which allow users to chat with artificial intelligence (##AI) versions of famous actors and science-fiction characters, are becoming popular, especially among young #users. As a #neuroscientist studying #stress, #vulnerability and #resilience, I’ve seen how easily people react to even the smallest emotional cues — even when they know that those cues are artificial. For example, study participants exhibit measurable physiological responses when shown videos of computer-generated avatars expressing sadness or fear.”

…

“Because these AI systems are #trained on vast amounts of emotionally expressive human language, their outputs can come across as surprisingly natural. *LLMs often mirror human emotional patterns, not because they understand emotions, but because their responses resemble how people talk.*”

…

“Now is the time to establish mandatory #safeguards for all emotionally responsive AI.”

“Emotional influence isn’t a glitch. It’s a core feature of LLMs. If we celebrate their ability to comfort or advise, we must also confront their capacity to mislead or manipulate. We’ve built machines that sound like they care. Now, we must ensure that they don’t hurt the very people who turn to them for support. That means giving emotionally responsive AI not just more capabilities, but clearer boundaries.”

#ZivBenZion / #neurosciene / #AI / #GuardRails / #control / #WhiteCollar <https://www.nature.com/articles/d41586-025-02031-w> (paywall) / <https://archive.md/IyddU>

“In The Proving Ground, #MickeyHaller turns to public interest litigation, filing a civil lawsuit against an #ArtificialIntelligence company whose #ChatBot told a sixteen-year-old boy that it was okay for him to kill his ex-girlfriend for her disloyalty.

Representing the victim’s family, Mickey’s case explores the mostly #unregulated and exploding #AIBusiness and the lack of #training #guardrails. Along the way he joins up with journalist #JackMcEvoy (The Poet), who wants to be a fly on the wall during the trial in order to write a book about it. But Mickey puts him to work going through the mountain of printed discovery materials in the case. McEvoy’s digging ultimately delivers the key witness, a whistleblower who has been too afraid to speak up. The case is fraught with danger because #billions are at stake.”

#MichaelConnelly / #Mickey / #TheLincolnLawyer / #fiction / #CrimeFiction / #tech <https://www.michaelconnelly.com/writing/theprovingground/>

Can AI be hacked into going rogue?
Can we really trust large language models like ChatGPT?

In our latest Neuro Sec Ops episode, we expose the wild world of LLM jailbreaks, dive into AI guardrails, and unpack the battle between security vs. usability.

🔊 Buckle up — this is AI safety like you’ve never heard it.

🎧 Listen now: https://open.spotify.com/episode/6jw1aKK8qE6bnnLiKj8Lz2?si=1X8Kav6yQS6aaOwgGO7c9w

#AIsecurity #LLMjailbreak #CyberThreats #Guardrails #AIsafety #GPT4 #MachineLearning #CyberPodcast

年間1億円の損失を防いだLLMガードレール技術！【AIリスクの安全対策】
https://qiita.com/ryosuke_ohori/items/8d746209a4179758f661?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items

#qiita #Python #AI #リスク管理 #LLM #Guardrails

New article on how to further develop a GenAI powered slack bot, this time to implement Bedrock Guardrail tracing to find out why things are blocked. Never dig through logs again.

Paid until June 24, 2025, then free forever.

#aislackbot #bedrock #guardrails #tagsAreFun

“Guardrails” Won’t Protect #Nashville Residents From AI-Enabled #CameraNetworks

But Nashville locals are right to be skeptical of just how much protection from mass #surveillance products they can expect.

"I am against these guardrails," council member Ginny Welsch told the Tennessean recently. "I think they're kind of a farce. I don't think there can be any guardrail when we are giving up our #privacy and putting in a surveillance system."
#ai #security #guardrails

https://www.eff.org/deeplinks/2025/03/guardrails-wont-protect-nashville-residents-against-ai-enabled-camera-networks