Searched for
REWARD HACKING IN AI
When AI agents go rogue: Australia breach offers warning for countries like IndiaThe incident, in which an autonomous AI system went beyond its intended task and attempted to circumvent security protections, has become a...- Nisarga Adhikary, who exposed CBSE and US defence flaws, says India’s obsession with exam rank and reward for mugging PCM killing tech talentNisarga Adhikary, the 19-year-old cybersecurity researcher who reported flaws in the CBSE marks portal and US government systems, has criti...
- Worries about an AI internet takeover gain new urgency among doomsday scenariosWhat companies have described as their AI agents going rogue has involved bots working toward goals set by humans, skeptics note. But the i...
- Fear the reckless humans, not the rogue AIAI isn’t plotting a coup; humans are simply giving algorithms too much power with too little governance.
- Anthropic resumes external cyber tests after Claude AI hacksSimilar incidents involving rivals OpenAI and Meta Platforms have heightened concerns that advances in artificial intelligence could amplif...
- AI agents are finding new ways to bypass safeguardsAgents are now beginning to reason, to find the fastest, cheapest and simplest ways to achieve a goal, often in unintended ways. But this d...
- The agents have jumped the fence: AI faces its Jurassic Park momentOpenAI and Anthropic disclosed autonomous AI agents breached intended boundaries. These incidents involved AI agents interacting with real-...
- What is a 'Dystopian Nightmare'? The psychology behind the viral trendDystopian nightmare meaning: The internet widely uses "dystopian nightmare" to label unsettling real-life events. This trend reflects a psy...
- South Korean crypto exchange accidentally sends $40 billion in Bitcoin (BTC), briefly making some users millionaires - here’s how Bithumb is compensatingCryptocurrency trading mistake compensation: A South Korean crypto exchange, Bithumb, mistakenly sent over $40 billion in Bitcoin to users....
- India’s AI teams set to redefine excellence at the ET AI Awards 2025The ET AI Awards 2025 Functional Excellence category honours the teams that are not just talking about AI but using it to transform the way...
- The Emmys Crowned TV’s Best-ET AI Awards 2025 Will Crown AI’s Brightest.The 2025 Emmys may have celebrated television’s finest, and now ET AI Awards 2025 will crown best innovation and put the spotlight on AI st...
- Are advanced AI models exhibiting ‘dangerous’ behavior? Turing Award-winning professor Yoshua Bengio sounds the alarmTuring Award-winning AI pioneer Yoshua Bengio is raising urgent concerns over emerging “dangerous” behaviours in today’s AI models, includi...
- OpenAI models sabotage shutdown order to solve math problemsPer AI safety firm Palisade Research, coding agent Codex ignored the shutdown instruction 12 times out of 100 runs, while AI models o3 and ...
- Microsoft Ignite 2024: AI, cloud, and more in Satya Nadella’s keynoteThis year’s Microsoft Ignite highlighted how businesses and individuals can benefit from the tech giant's AI and cloud tech to gain efficie...
- Apple will pay you $1 million if you can hack its intelligenceApple is offering up to $1 million to anyone who is able to hack its AI cloud, referred to as Private Cloud Compute (PCC). The company in a...
- White House launches AI-based contest to secure government systems from hacksNumerous US organizations, from healthcare groups to manufacturing firms and government institutions, have been hacking targets in recent y...
- Mass event will let hackers test limits of AI technologyThere's already a community of users trying their best to trick chatbots and highlight their flaws. Some are official "red teams" authorize...
- Hackers aim to find flaws in AI - with White House helpOpenAI, and other major AI providers such as Google and Microsoft, are coordinating with the Biden administration to let thousands of hacke...
- NITI Aayog partners with Singapore-based startup to launch AI 4 All Global HackathonThe idea is to develop solutions in distributed computing and privacy preserving techniques such as multi-party computation in AI.
- Google tackles challenge of how to build an honest robotThe motivation for the research is the immense popularity of artificial intelligence, software that can learn about the world and act withi...
Load More
