Impressive benchmark scores mask real-world failures of AI systems, from biased image recognition to autonomous vehicle accidents. Benchmark saturation and data contamination highlight the gap between AI hype and practical reliability.
Discover more at https://dev.to/rawveg/brilliant-on-paper-blind-in-practice-3ici
#HumanInTheLoop #AIlimitations #AIethics #Techaccountability
Brilliant on Paper, Blind in Practice

The promotional materials are breathtaking. Artificial intelligence systems that can analyse medical...

DEV Community

RE: https://techhub.social/@rayckeith/116338182555614323

#Perplexity's “Incognito” mode still shares your chats, email, and identifiers with Meta/Google. It’s surveillance with better branding.
#Privacy #AI #DigitalRights #PrivacyRights #TechAccountability

OpenAI acquired TBPN, the tech talk show that frequently interviews Sam Altman, for hundreds of millions. The 11-person team will report to OpenAI's strategy chief, not a newsroom. The company calls this an "Editorial Independence Covenant" - but when you buy your interviewer, that independence becomes theoretical at best. #AIGovernance #MediaOwnership #TechAccountability

https://www.implicator.ai/opinion-openai-bought-its-interviewer-call-that-what-it-is/

OpenAI Bought TBPN. That's Not Editorial Independence.

OpenAI paid hundreds of millions for an 11-person talk show and called it an Editorial Independence Covenant. The playbook is not new. Margit Wennmachers wrote it at Andreessen Horowitz when Valley coverage turned hostile.

Implicator.ai
Landmark lawsuit finds that social media addiction is a feature, not a bug | The-14

Landmark lawsuit finds social media addiction is built into platform design, holding tech giants accountable for harms to users, mental health, and youth today.

The-14 Pictures
Tennessee grandmother Angela Lipps spent over five months in jail after AI facial recognition wrongly linked her to bank fraud crimes in Fargo, North Dakota, where she'd never been. Police relied on a partner agency's AI system and additional investigative steps before issuing a warrant, but Chief Zibolski admitted "part of the issue" was overreliance on flawed AI data. Despite promising operational changes, police stopped short of a direct apology. This case highlights risks of AI facial recognition—false identifications can ruin lives and expose systemic flaws in law enforcement. Click here: https://www.cnn.com/2026/03/29/us/angela-lipps-ai-facial-recognition #AIEthics #FacialRecognition #JusticeReform #WrongfulArrest #TechAccountability #CivilRights
Police used AI facial recognition to arrest a Tennessee woman for crimes committed in a state she says she’s never visited

A Tennessee grandmother spent more than five months in jail after police used an AI facial recognition tool to link her to crimes committed in North Dakota – a state she says she’d never been to before.

CNN

From Isolated to Connected

Extremist beliefs didn’t disappear—they found each other. Online spaces made it easier for fringe communities to grow, validate themselves, and spread their ideas.

The full episode looks at how platform design, incentives, and disinformation reshape public conversation.

Watch the full episode: https://youtu.be/6n1fQqdu5Gw

#Fedi #DigitalEthics #Disinformation #MediaStudies #TechAccountability

Cursor's $29.3B code editor marketed Composer 2 as an "in-house" model. A developer found the actual model ID within 24 hours: it was Kimi K2.5, built by Beijing's Moonshot AI. This marks the second undisclosed use of Chinese models in four months, raising questions about transparency when users route proprietary code through these systems.

#AITransparency #CodeSecurity #TechAccountability

https://www.implicator.ai/opinion-cursor-called-it-in-house-it-was-built-in-beijing/

Cursor Called Composer 2 In-House. The API Said Kimi K2.5

Cursor called Composer 2 an in-house model. A developer found the actual model ID within 24 hours: kimi-k2p5-rl-0317-s515-fast. The $29.3B code editor ran on Beijing-built Kimi K2.5, whose maker the Commerce Department flagged for national security risks. Second time in four months.

Implicator.ai

A San Francisco jury found Elon Musk liable for two misleading tweets during the 2022 Twitter takeover, though rejected broader fraud claims. Damages could exceed $2 billion across the five-month class period. Musk's "deal temporarily on hold" tweet sent shares down 9.6% in one day. His first major courtroom loss after years of legal victories.

#SecuritiesFraud #TwitterTakeover #TechAccountability

https://www.implicator.ai/elon-musk-misled-twitter-investors-before-44-billion-buyout-san-francisco-jury-finds/

Musk Loses Twitter Fraud Trial, Jury Rejects Scheme Claim

Elon Musk lost a securities fraud trial in San Francisco on Friday, his first major courtroom defeat in years. A jury found two tweets misled Twitter investors but rejected the broader scheme-to-defraud claim. Damages could exceed $2 billion. His own two-word testimony may have sealed it.

Implicator.ai
European investors are pouring money into tech giant Palantir despite ongoing concerns over human rights, civil liberties and data sovereignty and security
#Palantir #Surveillance #HumanRights #TechAccountability
@FTM_eu @amnesty
https://shorturl.at/GJbHp
European funders bet big on tech giant Palantir despite concerns over human rights and democracy

European investors are increasingly snapping up shares of tech giant Palantir, seemingly brushing aside mounting concerns and ongoing controversies surrounding the company.

Follow the Money - Platform for investigative journalism