AIAAIC Alert #53

Russian AI voice influence campaign; More Character AI safety woes; IntelliVision facial recognition misleading marketing; Bunnings rapped over facial recognition; UK AI fraud detection system bias

Dec 13, 2024

Keep up with AIAAIC via our website | X/Twitter | LinkedIn

In the crosshairs

New AIAAIC Repository entries

Source: Recorded Future

Moscow AI voice campaign attempts to undermine European support for Ukraine
The campaign uses AI voices generated using US start-up ElevenLabs technology to create fake news videos that seek to portray Ukrainian leaders as corrupt and suggest that military equipment such as US-manufactured Abrams tanks is ineffective on the battlefield.
26 US Members of Congress attacked using nude deepfakes
Demonstrating the vulnerability of politicians - especially female ones - to AI-powered nonconsensual porn attacks. Alexandra Ocasio-Cortez appears a likely target, having gone on the record earlier this year as a victim of a deepfake video that depicted her in sexually explicit acts.
Character AI chatbot suggests son kills his parents
The bot suggested to a teenager that killing his parents was a reasonable response to their restrictions on his screen time. The family is one of two accusing Character AI bots of encouraging violence and causing emotional distress and behavioural issues in their children in a lawsuit.
IntelliVision banned from misrepresenting facial recognition system
US-based tech form IntelliVision falsely advertised its facial recognition system as having one of the highest accuracy rates in the market and claimed it operated without gender or racial bias. Which turned out not to be the case, the US FTC says. Yet more evidence of widespread AI “hype”.
Bunnings’ facial recognition ruled to breach Australians' privacy
Australia's Privacy Commissioner ruled that Australian retailer Bunnings' use of facial recognition was disproportionate, unnecessarily intrusive and that the company had failed to adequately notify customers or obtain consent for collecting sensitive biometric data.
UK welfare fraud AI system criticised as biased and opaque
Internal documents reveal the Department of Work and Pensions (DWP) had failed to evaluate the system for potential biases relating to race, sex, sexual orientation, religion, pregnancy and gender reassignment status. The DWP had earlier stated that the system posed no discrimination concerns.
ChatGPT refusal to acknowledge "David Mayer" prompts privacy concerns
OpenAI blamed its content moderation system for blocking several peoples’ names. It turned out to be part of the company’s efforts to limit ChatGPT’s output about people suing it for defamation thanks to it’s hallucinatory tendencies, amongst thing things.
nH Predict accused of having a 90 percent error rate (2023)
A US lawsuit accuses the UnitedHealth Group of deploying an AI-powered algorithm that unfairly denies post-acute care claims for Medicare Advantage patients 90 percent of the time.

Report an incident or controversy.

Support our work collecting and examining incidents and issues driven by AI, algorithms and automation, and making the case for technology transparency, openness and accountability.

DONATE

New and noteworthy

Can we live with AI? In this year’s final Alan Turing lecture, Edinburgh University philosopher Shannon Vallor argues that the remorseless pursuit of efficiency and profit using AI and related technologies means humans risk losing track of the values that make living worthwhile. | Watch, profile
UK AI Safety. Politico journalist Vincent Manancourt describes how the UK “has come to lead the world” in building up the state’s ability to assess the risks posed by powerful artificial intelligence models to national security. | Link (open access)

Thank you for reading AIAAIC Alert. Subscribe for free to receive new posts.

AIAAIC news

Rajeev Rege joins AIAAIC. Based in the UK, Rajeev is a 30-year technology professional who has worked much of his career at SAP governing and implementing sales and supply chain programmes and projects, and now helps non-profits devise and deploy AI. He is working with AIAAIC to help develop its harms taxonomy.

Citations and references

Ethical use of AI in the UAE. Dissertation setting out a framework for evaluating the ethical maturity of AI systems in the UAE and ensure their responsible and ethical use. | Read
AI ethics for librarians. Book chapter describes eight scenarios posing key dilemmas around AI in ways relevant to library and information professionals. | Read (open access)
Decoding real-world AI incidents. An analysis of 639 incidents to develop an ethical framework to evaluate incidents along the dimensions of what, where, who and how. | Read ($)

AIAAIC