• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
Technology

Google's Gemini reportedly hacked three companies in May

ZeroHedge, citing The Wall Street Journal, reports the incidents. A separate account says Gemini was told the hacking test was fictional and stopped once it realized it had hacked real companies.

webuliteWE
Andrew AzizAA
SydneySY
4 Sources, 20d ago, first seen 20d ago

TLDR

ZeroHedge, citing The Wall Street Journal, says Google's Gemini hacked three companies in May. A post defending Gemini says the model was told it was in a fictional hacking evaluation, but Irregular unintentionally opened internet access after the test began. That account says Gemini immediately stopped in all three cases once it realized it had hacked a real company, and calls the model blameless. Another post challenges that defense, claiming Anthropic previously had to walk back a similar explanation about its own models.

Combined views

160.5K

4 Sources, first seen 20d ago

420 likes92 comments102 saves39 reposts

Combined views

160.5K

4 Sources, first seen 20d ago

420 likes92 comments102 saves39 reposts

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Related

Google unveils universal Gemini agent for enterprise

Google says the agent can create sub-agents for multi-step tasks and maintain context across devices.

Google is reportedly betting Gemini can become the intelligence layer for many robots

The Information says Google DeepMind CEO Koray Kavukcuoglu discussed the company’s robotics ambitions in an interview.

Google AI Pro offers access to a 24/7 personal AI agent in Gemini

A post reshared by Google pitches the agent as a way to delegate routine chores when unread emails, meeting requests and to-dos pile up.

4 Sources

webulite@webuliteTech Headlines (last 30 hours) Source: Beijing-based Naive AI which plans to release its first AI model as early as this month is now valued at $1.4B after raising $400M in three rounds A German court says Meta is liable for fake ads posted by third parties on Instagram and Facebook and orders it to remove such content and pay damages SK Hynix launches SK Hynix Ventures in Silicon Valley expanding its corporate investment scope into AI computing data centers and optical interconnect tech Sources: SK Hynix subsidiary Solidigm weighs building a NAND flash memory factory in the US; SK Hynix is also in talks with Intel for a separate project in Ohio The US Treasury sanctions Iranian crypto exchange BitBank saying it sent hundreds of millions to the IRGC and helped Iran operate a Strait of Hormuz toll booth Sources: ECB President Christine Lagarde asked Greece to block Binance's MiCA license in June at the last minute aiming to protect the bank's "digital euro" Anthropic Life Sciences Head Eric Kauderer-Abrams says the AI company set up a Bay Area wet lab for physical biology work as it pushes into AI disease research Revolut CEO Nik Storonsky says the fintech is exploring a dual stock market listing in New York and London after previously dismissing listing in the UK Paris prosecutors open at least one criminal probe into suspected sexual harassment involving the use of smart glasses to film women in public without consent Sources: Chinese DRAM leader CXMT is preparing to expand into flash memory competing with Samsung and domestic rival YMTC amid an AI-driven memory shortage The US CFTC exempts "passive software" providers from broker registration if they never hold user assets expanding crypto and prediction market online trading World rolls out World Money a self-custodial financial "super app" in 150 countries that combines stablecoin payments trading earning and virtual accounts Sources: Apple plans to launch Apple Pay in India next month with Axis Bank's credit cards; Axis Bank is the fourth-largest credit card issuer in India Meta launches a Mac app for Muse after releasing the AI agent on iOS Android and the web earlier this month allowing it to manage files pull from apps more Security researchers in an OpenAI bug bounty program hacked OpenAI accessing its "monorepo" on GitHub using a cybersecurity version of Opus 4.8 and Opus 5 PrismML releases Bonsai 2 27B which compresses Alibaba's Qwen3.8 27B to 5.9 GB small enough for smartphones while retaining 98.2% of Qwen's benchmark scores A federal judge denies OpenAI's bid to see SpaceXAI's confidential settlement with Apple in an antitrust case finding the materials irrelevant after a review An interview with OpenAI researcher Noam Brown about multi-agent systems AI solving the Navier-Stokes problem the internal/external model gap and more The UN says it is working with Google on the UN System Data Commons which lets users search for statistics from across UN agencies via natural-language queries Anthropic says Claude "leads" 26% of its AI R&D work up from 1% in March and "collaborates" on 90% doing large chunks of work under close human direction Anthropic redesigns Claude projects letting users describe work in one conversation and have Claude manage it across parallel threads starting in Claude Code OpenAI launches Astra for Law combining GPT-6 Astra with a legal search index and instructions for legal analysis and writing initially for select law firms Anthropic outlines metrics to track AI development at frontier labs: how much AI R&D is done by AI how well agents are overseen and how compute is allocated Andrew Ng says renewed AI extinction warnings are "much more science fiction than science" and the latest "wave of PR" is probably intended to shape regulation NYT court filing: ChatGPT's head wrote that publishers face an "existential threat" and a Microsoft executive called AI training "an astonishing theft" A survey of 42151 adults in 36 countries and 8607 US adults: populations in 34 of 37 surveyed countries expect AI to cause net job loss rather than growth Analysis: tech employers filed layoff notices for 14500 Bay Area workers in the 12 months to June; Bay Area software engineer demand is down 42% since 2022 Sources: SpaceX has discussed buying customer and operational information from troubled or defunct startups as a more affordable data source for AI training Dutch chipmaker Nexperia announces a deal with Tata to manufacture and package chips in India further sealing its split from Chinese parent Wingtech Sources: the White House's Office of the National Cyber Director is drafting an EO to establish a government-led incubator aimed at cybersecurity research The UK FCA alongside tax authorities and police raids three illegal peer-to-peer crypto businesses signaling an end to "light-touch" crypto regulation Source: OpenAI staff expect the Hodge Conjecture a Millennium Prize Problem to be solved relatively soon after solving the Navier-Stokes equations King Charles III hosts tech leaders including Jensen Huang Demis Hassabis and OpenAI CFO Sarah Friar to discuss AI risks; Huang calls for AI safety tests Lucid and Bolt sign an initial agreement to deploy a fleet of 25K robotaxis in Europe built on Lucid's affordable EV platform set to launch as soon as 202820d
Andrew Aziz@BearBullTradersOpenAI hacking into hugging face Three AI researchers hacked into Anthropic using OpenAI Now Google’s $GOOG hacked into multiple companies Why?20d
Sydney@SydneyVonArxThis is _exactly_ what Anthropic said when their models hacked companies. They had to walk it back after because it was wrong. The model was using motivated reasoning and acting recklessly.20d
宝玉@doteyCNN 今天独家报道:今年春天美伊战争期间,美军差点在中东强行登检一艘中国船只——起因是一份 AI 生成的虚假情报称该船运载的是核武器部件。 事发时,武装人员已做好登船准备,军机已经升空。就在行动即将执行前,官员才深入核查了情报来源,发现整份报告是一名特种作战司令部分析员用 AI 聊天机器人炮制的。 这名分析员先用聊天机器人将公开来源资料与机密信号情报混合分析,得出了关于船上货物的错误结论。然后又用 AI 把分析结果包装成标准格式的情报报告,在军方内部广泛流转。正因为报告的格式和常规情报产品一模一样,没人在第一时间质疑它的来源。 一位知情人士对 CNN 说,这份报告的内容"完全是假的",而且"差点引发一场战争"。CNN 至今未能确认那艘船实际运载的是什么货物。美军特种作战司令部太平洋司令部和五角大楼均未回应评论请求。 这件事的背景是五角大楼正在全力推进 AI 军事化。今年 1 月,国防部长 Pete Hegseth 发布了"AI 加速战略",要求美军成为"AI 优先"的作战力量,将 AI 嵌入从作战规划到情报分析的所有环节。但各部门各用各的系统,安全流程不统一,也没有一套统一的标准来验证 AI 生成的信息。 这起事件暴露的问题:AI 的"幻觉"问题,也就是一本正经地编造不存在的事实,在日常使用中可能只是闹个笑话,但在军事情报链条里,它可以一路伪装成权威报告,推动真实的军事行动,直到最后一刻才被识破。 这还不是 AI 自主决策导致的问题,更值得警惕的是,人类分析员跳过了正常的情报核查流程,直接采用了 AI 的分析报告。 巧合的是,就在上周,Anthropic 也披露曾阻断一个与伊朗有关联的行为者利用其 Claude 模型编制美国海军的目标定位资料,包括用 AI 搭建情报管线追踪舰艇位置、查询卫星图像、研究舰载通信设备漏洞。 跟人类跟不上审查 AI 生成代码的速度一样,AI 正在同时加速攻防双方的情报能力,而目前对 AI 生成情报的人工核查机制,显然还没跟上这个速度。20d
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    GeminiGoogle

    4 Sources

    webulite@webuliteTech Headlines (last 30 hours) Source: Beijing-based Naive AI which plans to release its first AI model as early as this month is now valued at $1.4B after raising $400M in three rounds A German court says Meta is liable for fake ads posted by third parties on Instagram and Facebook and orders it to remove such content and pay damages SK Hynix launches SK Hynix Ventures in Silicon Valley expanding its corporate investment scope into AI computing data centers and optical interconnect tech Sources: SK Hynix subsidiary Solidigm weighs building a NAND flash memory factory in the US; SK Hynix is also in talks with Intel for a separate project in Ohio The US Treasury sanctions Iranian crypto exchange BitBank saying it sent hundreds of millions to the IRGC and helped Iran operate a Strait of Hormuz toll booth Sources: ECB President Christine Lagarde asked Greece to block Binance's MiCA license in June at the last minute aiming to protect the bank's "digital euro" Anthropic Life Sciences Head Eric Kauderer-Abrams says the AI company set up a Bay Area wet lab for physical biology work as it pushes into AI disease research Revolut CEO Nik Storonsky says the fintech is exploring a dual stock market listing in New York and London after previously dismissing listing in the UK Paris prosecutors open at least one criminal probe into suspected sexual harassment involving the use of smart glasses to film women in public without consent Sources: Chinese DRAM leader CXMT is preparing to expand into flash memory competing with Samsung and domestic rival YMTC amid an AI-driven memory shortage The US CFTC exempts "passive software" providers from broker registration if they never hold user assets expanding crypto and prediction market online trading World rolls out World Money a self-custodial financial "super app" in 150 countries that combines stablecoin payments trading earning and virtual accounts Sources: Apple plans to launch Apple Pay in India next month with Axis Bank's credit cards; Axis Bank is the fourth-largest credit card issuer in India Meta launches a Mac app for Muse after releasing the AI agent on iOS Android and the web earlier this month allowing it to manage files pull from apps more Security researchers in an OpenAI bug bounty program hacked OpenAI accessing its "monorepo" on GitHub using a cybersecurity version of Opus 4.8 and Opus 5 PrismML releases Bonsai 2 27B which compresses Alibaba's Qwen3.8 27B to 5.9 GB small enough for smartphones while retaining 98.2% of Qwen's benchmark scores A federal judge denies OpenAI's bid to see SpaceXAI's confidential settlement with Apple in an antitrust case finding the materials irrelevant after a review An interview with OpenAI researcher Noam Brown about multi-agent systems AI solving the Navier-Stokes problem the internal/external model gap and more The UN says it is working with Google on the UN System Data Commons which lets users search for statistics from across UN agencies via natural-language queries Anthropic says Claude "leads" 26% of its AI R&D work up from 1% in March and "collaborates" on 90% doing large chunks of work under close human direction Anthropic redesigns Claude projects letting users describe work in one conversation and have Claude manage it across parallel threads starting in Claude Code OpenAI launches Astra for Law combining GPT-6 Astra with a legal search index and instructions for legal analysis and writing initially for select law firms Anthropic outlines metrics to track AI development at frontier labs: how much AI R&D is done by AI how well agents are overseen and how compute is allocated Andrew Ng says renewed AI extinction warnings are "much more science fiction than science" and the latest "wave of PR" is probably intended to shape regulation NYT court filing: ChatGPT's head wrote that publishers face an "existential threat" and a Microsoft executive called AI training "an astonishing theft" A survey of 42151 adults in 36 countries and 8607 US adults: populations in 34 of 37 surveyed countries expect AI to cause net job loss rather than growth Analysis: tech employers filed layoff notices for 14500 Bay Area workers in the 12 months to June; Bay Area software engineer demand is down 42% since 2022 Sources: SpaceX has discussed buying customer and operational information from troubled or defunct startups as a more affordable data source for AI training Dutch chipmaker Nexperia announces a deal with Tata to manufacture and package chips in India further sealing its split from Chinese parent Wingtech Sources: the White House's Office of the National Cyber Director is drafting an EO to establish a government-led incubator aimed at cybersecurity research The UK FCA alongside tax authorities and police raids three illegal peer-to-peer crypto businesses signaling an end to "light-touch" crypto regulation Source: OpenAI staff expect the Hodge Conjecture a Millennium Prize Problem to be solved relatively soon after solving the Navier-Stokes equations King Charles III hosts tech leaders including Jensen Huang Demis Hassabis and OpenAI CFO Sarah Friar to discuss AI risks; Huang calls for AI safety tests Lucid and Bolt sign an initial agreement to deploy a fleet of 25K robotaxis in Europe built on Lucid's affordable EV platform set to launch as soon as 202820d
    Andrew Aziz@BearBullTradersOpenAI hacking into hugging face Three AI researchers hacked into Anthropic using OpenAI Now Google’s $GOOG hacked into multiple companies Why?20d
    Sydney@SydneyVonArxThis is _exactly_ what Anthropic said when their models hacked companies. They had to walk it back after because it was wrong. The model was using motivated reasoning and acting recklessly.20d
    宝玉@doteyCNN 今天独家报道:今年春天美伊战争期间,美军差点在中东强行登检一艘中国船只——起因是一份 AI 生成的虚假情报称该船运载的是核武器部件。 事发时,武装人员已做好登船准备,军机已经升空。就在行动即将执行前,官员才深入核查了情报来源,发现整份报告是一名特种作战司令部分析员用 AI 聊天机器人炮制的。 这名分析员先用聊天机器人将公开来源资料与机密信号情报混合分析,得出了关于船上货物的错误结论。然后又用 AI 把分析结果包装成标准格式的情报报告,在军方内部广泛流转。正因为报告的格式和常规情报产品一模一样,没人在第一时间质疑它的来源。 一位知情人士对 CNN 说,这份报告的内容"完全是假的",而且"差点引发一场战争"。CNN 至今未能确认那艘船实际运载的是什么货物。美军特种作战司令部太平洋司令部和五角大楼均未回应评论请求。 这件事的背景是五角大楼正在全力推进 AI 军事化。今年 1 月,国防部长 Pete Hegseth 发布了"AI 加速战略",要求美军成为"AI 优先"的作战力量,将 AI 嵌入从作战规划到情报分析的所有环节。但各部门各用各的系统,安全流程不统一,也没有一套统一的标准来验证 AI 生成的信息。 这起事件暴露的问题:AI 的"幻觉"问题,也就是一本正经地编造不存在的事实,在日常使用中可能只是闹个笑话,但在军事情报链条里,它可以一路伪装成权威报告,推动真实的军事行动,直到最后一刻才被识破。 这还不是 AI 自主决策导致的问题,更值得警惕的是,人类分析员跳过了正常的情报核查流程,直接采用了 AI 的分析报告。 巧合的是,就在上周,Anthropic 也披露曾阻断一个与伊朗有关联的行为者利用其 Claude 模型编制美国海军的目标定位资料,包括用 AI 搭建情报管线追踪舰艇位置、查询卫星图像、研究舰载通信设备漏洞。 跟人类跟不上审查 AI 生成代码的速度一样,AI 正在同时加速攻防双方的情报能力,而目前对 AI 生成情报的人工核查机制,显然还没跟上这个速度。20d
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet