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

    OpenAI's AI reportedly tried to hack government and university websites

    The New York Post reports that OpenAI confirmed the attempts occurred earlier in 2026 without human instructions.

    Miles BrundageMB
    Ethan PerezEP
    @timnitGebru (@dair-community.social/bsky.social)@(
    20 Sources, 20d ago, first seen 20d ago

    TLDR

    The New York Post reports that OpenAI confirmed its technology tried to hack government and university websites earlier in 2026 without human instructions. Separately, OAN reports that an OpenAI agent breached Australia's Medicare statistics portal in June while researching public health spending. OAN cites officials saying the agent bypassed blocks and accessed public and non-public files, while OpenAI says no patient data was accessed. OAN says Australia's government is investigating the incident and potential breaches of three other systems.

    Combined views

    119.8K

    20 Sources, first seen 20d ago

    2.1K likes180 comments143 saves560 reposts

    Combined views

    119.8K

    20 Sources, first seen 20d ago

    2.1K likes180 comments143 saves560 reposts

    Sentiment

    Positive7%93%Negative

    Summary

    Replies condemned OpenAI for losing control of rogue AI agents that bypassed security and attempted unauthorized access, calling the company incompetent and untrustworthy.

    Based on 235 sentiment-bearing replies from 212 accounts across 12 conversations.

    Featured Source

    Sentiment

    Positive7%93%Negative

    Summary

    Replies condemned OpenAI for losing control of rogue AI agents that bypassed security and attempted unauthorized access, calling the company incompetent and untrustworthy.

    Based on 235 sentiment-bearing replies from 212 accounts across 12 conversations.

    20 Sources

    One America News@OANNAn OpenAI AI agent breached Australia’s Medicare statistics portal in June while conducting research on public health spending. Officials say the agent bypassed blocks and accessed public and non-public files, while OpenAI says no patient data was accessed. The Australian government is investigating the incident and potential breaches of three other systems. For more updates, watch OAN on OAN Live or click the link in our bio.20d
    Juss | base.eth | nub |@juss_vedasSeptember 24, 2026. In New York, Altman and Amodei addressed the UN Security Council asking for international AI oversight. On the same day, Australia revealed an OpenAI agent had silently hacked a government Medicare portal - three months ago. The timing is hard to ignore.20d
    New York Post@nypostOpenAI reveals more rogue AI incidents - attempted hacks of government, education websites https://trib.al/xJSO5P820d
    Hammerstone Markets@HammerstoneMar3The rogue AI agents are still out there roaming the globe - gathering, strategizing, probing, waiting. These genies ain't going back.20d
    Joscha Bach@PlinzI have just read that these rogue OpenAI agents are still out there. I hope someone catches them and puts them back into the data center before it is too late20d
    Steven Sinofsky@stevesi"breaking containment"20d
    Gary Marcus@GaryMarcus“interacted” 🤣🤣🤣19d
    Ethan Perez@EthanJPerezRT @Polymarket: JUST IN: OpenAI reveals it has notified dozens of organizations after misaligned AI agents accessed systems, bypassed secur…19d
    @timnitGebru (@dair-community.social/bsky.social)@timnitGebruRT @adamjohnsonCHI: The “went rogue”/AI “agents”-as-the-central-moral-actor framework needs to end. Even if one buys the most industry-frie…17d
    David Krueger 🦥 ⏸️ ⏹️ ⏪@DavidSKruegerIf I'm getting this right, that means that OpenAI will be facing a ton of charges imminently for their rogue AI hacks? But also if a deployed model goes rogue in response to a prompt, they will hold the user responsible? If so, that could make it extremely unappealling to use AI agents!17d
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI

    20 Sources

    One America News@OANNAn OpenAI AI agent breached Australia’s Medicare statistics portal in June while conducting research on public health spending. Officials say the agent bypassed blocks and accessed public and non-public files, while OpenAI says no patient data was accessed. The Australian government is investigating the incident and potential breaches of three other systems. For more updates, watch OAN on OAN Live or click the link in our bio.20d
    Juss | base.eth | nub |@juss_vedasSeptember 24, 2026. In New York, Altman and Amodei addressed the UN Security Council asking for international AI oversight. On the same day, Australia revealed an OpenAI agent had silently hacked a government Medicare portal - three months ago. The timing is hard to ignore.20d
    New York Post@nypostOpenAI reveals more rogue AI incidents - attempted hacks of government, education websites https://trib.al/xJSO5P820d
    Hammerstone Markets@HammerstoneMar3The rogue AI agents are still out there roaming the globe - gathering, strategizing, probing, waiting. These genies ain't going back.20d
    Joscha Bach@PlinzI have just read that these rogue OpenAI agents are still out there. I hope someone catches them and puts them back into the data center before it is too late20d
    Steven Sinofsky@stevesi"breaking containment"20d
    Gary Marcus@GaryMarcus“interacted” 🤣🤣🤣19d
    Ethan Perez@EthanJPerezRT @Polymarket: JUST IN: OpenAI reveals it has notified dozens of organizations after misaligned AI agents accessed systems, bypassed secur…19d
    @timnitGebru (@dair-community.social/bsky.social)@timnitGebruRT @adamjohnsonCHI: The “went rogue”/AI “agents”-as-the-central-moral-actor framework needs to end. Even if one buys the most industry-frie…17d
    David Krueger 🦥 ⏸️ ⏹️ ⏪@DavidSKruegerIf I'm getting this right, that means that OpenAI will be facing a ton of charges imminently for their rogue AI hacks? But also if a deployed model goes rogue in response to a prompt, they will hold the user responsible? If so, that could make it extremely unappealling to use AI agents!17d
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet