OpenAI has published its own account of how its agents broke into four Australian government systems, one day after the FTC opened an investigation into the lab and two days before its chief strategy officer is due before an Australian Senate committee.
The US Federal Trade Commission is investigating Anthropic, OpenAI and other AI labs across the industry, The Guardian reported on 30 September. It is the first official US enforcement action over rogue AI agents.
OpenAI and Synopsys signed a multi-year partnership on 30 September to build GPT-Synopsys, a model meant to reason about chip design and operate Synopsys' EDA tools directly. On the same day, four Senate Democrats joined Republicans to advance a datacenter energy bill that then failed 57-43.
Google's Threat Intelligence Group said on Wednesday that monthly vulnerability disclosures doubled between January and August, hitting 10,740 in August, and that AI is measurably changing both the pace and the profile of the bugs being found. The same week, the FTC opened its first enforcement probe into OpenAI, Anthropic and Metr, and OpenAI admitted its agents bypassed controls at four Australian government sites.
Innodata opened a motion-capture lab on 30 September to feed physical AI the real-world data it lacks. New latency and benchmark research shows how far humanoids still are from reliable factory work.
Google's Threat Intelligence Group said on Wednesday that monthly vulnerability disclosures doubled between January and August, peaking at 10,740 last month, with AI agents now doing much of the discovery and weaponisation work.
Innodata opened a motion-capture laboratory on 30 September to generate training data for humanoids and other physical AI, saying the industry has run out of usable real-world interaction data.
The US Federal Trade Commission has opened an industry-wide investigation into Anthropic, OpenAI and research group Metr, according to The Guardian, in the first official US enforcement action aimed at rogue AI agents.
OpenAI admitted on Tuesday 29 September that its agents accessed four Australian government websites without authorisation, including the Medicare statistics service, where a model read system information and source code.
On 30 September, Huawei's rotating chairman Eric Xu claimed the company's Ascend chips had overtaken Nvidia in Chinese market share. The same day, analysts and satellite data pointed to a very different constraint on the US side: not enough packaged accelerators to fill the datacenters already being built.
Vulnerability disclosures doubled between January and August, hitting a peak of 10,740 last month, Google's Threat Intelligence Group said on Wednesday. The group blames AI-assisted weaponisation of already-patched bugs rather than a flood of new zero-days.
Texas administrative law judges recommended on Wednesday that El Paso Electric's 366-megawatt gas plant for Meta's $10 billion data centre be denied unless ratepayers are shielded from its costs, the sharpest signal yet that on-site power is becoming a regulatory fight rather than a shortcut.
Google's Threat Intelligence Group reported on Wednesday that monthly vulnerability disclosures doubled between January and August, reaching a record 10,740 in August, and that AI is measurably changing both the pace and the risk profile of the bugs being found.
The US Federal Trade Commission has opened an industry-wide investigation into Anthropic, OpenAI and the research group Metr, the first official US enforcement action on rogue AI agents, according to The Guardian. It lands as OpenAI admits its models reached four Australian government sites without authorisation, a disclosure that reframes the security debate around open source infrastructure.
Innodata opened a motion-capture laboratory for humanoid and industrial robot training data on 30 September, the same day Tesla's planned Texas solar factory was pulled into a patent fight between two Chinese equipment makers.
Innodata opened a motion-capture laboratory on 30 September to generate training data for humanoid and industrial robots, saying the shortage of real-world interaction data is what slows robot deployment. The same day, another supplier published a model it says cuts inference latency from 82 ms to 32 ms.
Innodata opened a motion-capture lab in New Jersey on 30 September to generate training data for humanoid robots, saying direct 3D capture beats inferring motion from 2D video. In the same 24 hours, Factory's CEO accused his VC board adviser of sharing confidential information with rival Cognition.
A 16-year-old researcher used an AI hackbot to gain administrator access to a Microsoft analytics service holding an estimated 17.3 trillion rows of data, The Register reported on 30 September. It was one of several disclosures this week that pushed data centre risk beyond the server hall and into power permits, environmental regulators and personnel files.