AIAAIC

OpenAI's open source, multi-language automatic speech recognition model Whisper is widely used to provide real-time transcription for live events and is integrated into healthcare and a host of other sectors. Here's a dispassionate look at its benefits, risks and harms. https://lnkd.in/eiNsBx75 #AI #AIethics #OpenAI #Whisper

Issue 53 of our newsletter details the latest meltdowns and misuses of AI and algorithmic systems, notably: - A Russian AI-powered voice influence campaign used ElevenLabs technology to sow distrust across Europe - 26 Members of Congress - mostly women - were attacked with nude deepfakes, according to new research - A Character AI chatbot suggested a son kill his parents - US tech form IntelliVision, a Video/Audio AI/Analytics Company was rapped by the US FTC for using misleading marketing to describe its facial recognition system - An opaque UK Department for Work and Pensions (DWP) welfare fraud system was criticised as biased - Australian retailer Bunnings was rapped over its "disproportionate" and "unnecessarily intrusive" use of facial recognition. We are also publicly welcome Rajeev Rege to our ranks. Happy reading. https://lnkd.in/emVD7E6x #AI #AIethics

Issue 52 of our newsletter records the latest set of AI and algorithmic meltdowns and misuses: - UnitedHealth Group AI-powered algorithm is accused of systematically limiting mental health coverage - ChatGPT is found to misattribute and misrepresent news publisher content - Prominent Canadian news publishers sue OpenAI for alleged copyright abuse - Netherlands education fraud algorithm is ruled discriminatory and illegal - An applicant tracking system that auto-rejects qualified resumes results in an HR's team dismissal - Netflix is blasted for a "disrespectful" Arcane AI marketing poster Also highlighted is work by AIAAIC contributors Djalel Benbouzid and Arthit Suriyawongkul on LLM development challenges and an AI Bill of Materials. https://lnkd.in/e8p33vwM #AI #AIsafety #AIethics #ethics #chatgpt #OpenAI #UnitedHealth

Issue 52 of our newsletter details a slew of AI and algorithmic system misuses and meltdowns, including: - 3 Indians die after Google Maps directs them over an unfinished bridge -  Anthropic, Meta, Salesforce and other firms quietly appropriated 100,000+ TV and film scripts to train their AI models - A German music rights organisation sued OpenAI for abusing the copyright of 95,000 of its members - Grok was called out for mistaking sensitive medical data - "David Attenborough" was used to present war news reports - A Stanford professor was accused of using AI to write his court testimony supporting a ban on political deepfakes - A UK AI immigration enforcement tool was criticised as automated "rubberstamping" exercise - A marketing campaign to promote Japan's Fukuoka prefecture hallucinated attractions and cuisines - AI-powered shopping ads by Australian start-up Trendii appeared over photos of bomb destruction in Gaza and other gruesome stuff https://lnkd.in/egVDxkaD #AI #AIethics #GoogleMaps #Anthropic #Salesforce #OpenAI #Grok

Midjourney was early to the AI image generation game, and remains one of the better known tools. Here's a dispassionate look at its benefits, risks and harms. https://lnkd.in/eG7zMMnE #midjourney #AI #AIethics

The use and misuse of AI continues to inflict harm; the latest issue of our newsletter details incidents recorded over the last few days: - A Coke generative AI ad was criticised as soulless and for endangering human jobs - Amnesty accused Denmark's AI-powered welfare system of being an illegal social scoring system - ChatGPT singularly failed to debunk US election misinformation - AI-powered review platform Sitejabber was charged with deception - Canadian legal information database CanLII sued Caseway AI for copyright abuse - Google's Gemini chatbot implored a student to die - Dozens of suicide, paedophile chatbots were found, calling into question the sincerity and enforcement of Character AI's rules - A Lancaster Country School student attacked colleagues with nude deepfakes - Singapore students were humiliated with nude deepfakes https://lnkd.in/gNUFYdMX #AI #AIethics #AIsafety #chatgpt #google #characterAI

The latest issue our newsletter is out, highlighting a dramatic increase in e-waste caused by generative AI; ChatGPT's use to sway Ghana's political elections; LLMs passing personal info to hackers - and a host of other incidents and issues. In addition, we take a look at the pros and cons of MIT/NYU's famous/infamous 80 Million Tiny Images dataset, detail an Amnesty report on Denmark's potentially highly damaging welfare fraud detection system and the knock-on effects of data centres in and around Johore, Malaysia. Plenty to ponder over the weekend. https://lnkd.in/enhPhGSK #AI #AIethics #environment #ewaste #datacentre #malaysia #chatgpt #tinder #deepfake

Here's our latest newsletter, featuring latest incidents and a look at the pros and cons of Character.AI. Also welcoming Charlie Collins M.Eng to our ranks. Enjoy! https://lnkd.in/eq_UTQA4

Issue 45 of our newsletter takes a look at the pros and cons of Google's AI Overviews, details incidents involving Ray-Ban Meta smart glasses and Snap's failure to handle sextortion reports, and highlights reports on how to involve citizens in global AI governance and the failure of AI labs to have proper risk management systems. Have a great weekend. https://lnkd.in/equ9Ve55 #AI #AIethics #google #rayban #snapchat

Here's our latest newsletter, featuring notable AI and algorithmic backlashes over the past few days, a look at the controversial ClothOff nudifier and how French authorities are encouraging civil society participation in their AI Action Summit, and news of our new user guide. *Happy* reading. https://lnkd.in/ejDVvsfg #AI #AIethics #udemy #chatgpt #openai #facebook #climatechange #nudifier