AI chatbots amplify creation of false memories, boffins say - The Register
AI chatbots, known for their habit ofhallucinating, can induce people tohallucinatetoo, researchers claim. More specifically, interactions with …
The Register 60
Covers: Hallucinations & FabricationChatbots & LLMsRegulation & PolicyCybersecurityEnergy & EnvironmentAI Agents & AutomationGovernment & Public SectorLegal & CourtsJobs & the EconomyMilitary & Defense
EUAIAct in infancy, but using 'intelligent' HR apps arisk- The Register
The Register Some say that by introducing theAIAct as it stands, the European Union, the world's richest economic and political bloc,risksmissing out on that …
You're nothallucinating: generative AI is helping IBM's mainframes grow - The Register
Generative AI'spowers extend to helping the ancient concept of a proprietary enterprise OS and hardware stack to thrive, if IBM's Q2 2024 results …
Honey, I shrunk the LLM! A beginner's guide to quantization – and testing it - The Register
...hallucinations. About half the time, the 1.56-bit quantized Mistral ...AI: The first person to set foot on the moon was not a human, but …
White House urged to probe $1.5B G42-Microsoft AI deal - The Register
The two Republicans warn the Microsoft deal raises theriskof advanced American AI technology making its way to China via G42. "This deal may be …
FBI, cyber-cops zap 968 'Russian AI disinfo' Twitter bots - The Register
... that uses AI to make up fake data, and used to vocally spread political opinions; another archetype with very little info at all used for liking …
An introduction to embedding an LLM into your application • The Register - Theregister
In Rust, we trust. But in gen-AI to nothallucinate? Eh, that's another story · Hardware and software support · Grabbing our dependencies · Installing …
Meta warns that bit flips and other hardware faults cause AI errors - The Register
It's nohallucination: Roughly 4 in 1,000 inferences inaccurate due to this alone ... Meta has identified another reason AI might produce rubbish output …
Study uses large language models to sniff outhallucinations- The Register
The Register The Oxford team's approach used LLMs' ability to recognize entailment as a way of spotting confabulations in another LLM. World's top AI chatbots have …
US senators propose guardrails for government AI purchases and operations - The Register
...AI"while safeguarding against potentialrisksand harms." "Artificial intelligencehas the power to reshape how the federal government provides …
Italian premier and Pope to warn G7 of AI dangers- The Register
The Register Italian prime minister Giorgia Meloni and Pope Francis have joined forces to deliver a high-stakes warning to world leaders: Dive headfirst into …
Don't fall for the bring-your-own-AI trap - The Register
Bring-your-own-AI by any other name is shadow AI, and unsanctioned AI posesrisksto your business ...risks. Your AI strategy will be pivotal to your …
Developers seethe as Google surfaces buggyAI-written code - The Register.
While tools like Pulumi AI can provide value to some, filling the internet with unconfirmed, possiblyhallucinated, answers is actually pretty …
UK unions publish AI bill to protect workers from 'risksand harms' of tech - The Register.
“The Bill also provides for trade union rights in relation to the use ofartificial intelligencesystems by employers, addresses therisksassociated …
ChatGPT forecasts the future better when telling tales • The Register - Theregister
AI models become better at foretelling the future when asked to frame the prediction as a story about the past, boffins at Baylor University have …
Google Cloud chief is really psyched about this AI thing - The Register.
...hallucination, usage of resources like energy ... VertexAI, Google's enterprise AI...AI and Gen AI to life in enterprise data and execution contexts …
AI will reduce workforce, say 41% of execs in a survey - The Register.
AI researchers have started reviewing their peers using AI assistance ... Defenses against prompt injection, hallucination arrive as Feds eye MLrisks …
Jobs & the EconomyHallucinations & FabricationCybersecurityMilitary & Defense
AI will reduce workforce, say 41% of execs in a survey - The Register.
Not concerning themselves with thehallucinationsof generative AI and the refusal of vendors to take responsibility for the output of AI in the …
Hallucinations & FabricationJobs & the EconomyChatbots & LLMs
UK and US to jointly develop AI test suites to tacklerisks- The Register.
The Register....threatsposed by AI and the need to mitigate therisksassociated with the technology. The UK and USAISafety Institutes have laid out proposals …
Microsoft rolls out safety tools for Azure AI. Hint: More models - The Register.
Defenses against prompt injection,hallucinationarrive as Feds eye ML risks ... Microsoft has introduced a set of tools allegedly to help make AI …
Hallucinations & FabricationAI Safety & Existential RiskCybersecurityMilitary & Defense
Microsoft rolls out safety tools for Azure AI. Hint: More models - The Register.
Defenses against prompt injection, hallucination arrive as Feds eye MLrisks... Microsoft has introduced a set of tools allegedly to help make AI …
AI Safety & Existential RiskHallucinations & FabricationCybersecurityMilitary & Defense
AI botshallucinatesoftware packages and devs download them - The Register.
...hallucinatedby generative AI. Not only that but someone, having spotted this reoccurringhallucination, had turned that made-up dependency into a …
UN adopts US-sponsored, and non-binding,AI resolution - The Register
The United Nations has unanimously adopted a resolution aimed at establishing international AI development standards …
Large language model SauLM-7B aims for legal applications - The Register
The tendency of AI models tohallucinateand their uncertain data provenance would appear to be deal breakers in an industry where the stakes are …
Election securitythreatsin 2024 range from AI to … anthrax - The Register
State and county officials have been urged to use paper ballots wherever possible over electronic ones, and ensure all election offices have …
Mitigating AI securityrisks• The Register - Theregister
Because it's not just the good guys who useAI- bad actors power up their cyber attacks to bypass security measures and exploit vulnerabilities too …
California proposes government cloud cluster to sift out nasty AI models - Theregister
...risksstemming from increasingly powerful AI... The consultation in preparation for developing legislation that addressesrisksinherent in deploying …
NIST warns of 'snake oil' security claims by AI makers - The Register
...risksposed by AI systems. Overall, the results don't look good. The paper [PDF], titled, "Adversarial Machine Learning: A Taxonomy and …
EU agrees on Act that bans some AIs • The Register
AI systems deemed "high-risk" – such as those related to critical infrastructures, education access assessment, law enforcement, or biometric …
EU running in circles trying to getAIAct out the door - The Register
Blocrisksmissing out on first-to-legislate status if timetable slips · Forrester principal analyst Enza Iannopollo said: "The potential delay on the …