Evaluating Contract Bids With AI? Prepare to Be Challenged.

by  David Smalley

A $450M bid protest over an alleged AI hallucination is a preview of what's coming for every agency using AI in source selection, writes David Smalley with defense AI firm Seekr.

Summary

A federal contractor lawsuit over the Army’s use of AI to award a $450 million contract alleges the decision relied on “a classic AI hallucination,” a case that could reshape how agencies must document and disclose AI’s role in procurement decisions.

The piece argues that informing a decision and being held accountable for one are separate standards that AI deployment has largely failed to reconcile — AI outputs offer no visibility into what was weighted, discarded, or how confident the system was.

The author calls for four auditable capabilities: auditability, observability, contestability, and continuous evaluations, arguing agencies should be able to explain and defend any AI–influenced decision, at any time, before a regulator or court demands it.

 Read full story at Washington Technology

Lawmakers Call for Government Intervention After AI Leaders Issue Warning

by  Mandy Taheri & Hollie Silverman

Lawmakers are calling for federal oversight of artificial intelligence following warnings from industry leaders about the rapid advancement of AI technology.

Summary

Anthropic CEO Dario Amodei, along with other tech executives, cautioned that current safety measures are insufficient to keep pace with AI development.

Lawmakers are advocating for stronger regulatory frameworks, international cooperation, and a potential pause on the development of advanced AI systems to ensure safety and prevent misuse.

 Read full story at Newsweek

The US Military Gets Its Own ChatGPT

by  Patrick Tucker

After months of security testing, Pentagon adds an OpenAI product to the list of AI tools approved for unclassified work.

Summary

The Pentagon has added OpenAI’s ChatGPT Mil and Grok for Government to GenAI.mil, joining Google Gemini on the department’s generative-AI platform for controlled unclassified information. Officials said each model suits different tasks — Gemini for search and scraping, ChatGPT for text-based work — giving personnel flexibility to choose per task.

ChatGPT Mil, built since February, targets document-heavy work like planning, logistics, and administration across more than 3 million Department personnel. Like Gemini, it includes added security engineering separating it from the commercial version to safeguard CUI.

Officials from both OpenAI and the Pentagon said the tools are not intended for combat use, citing connectivity and computing demands as poor fits for that environment.

 Read full story at NextGov

Are USB-C Flash Drives Better Than USB-A Ones?

by  Thomas Godwin

USB-C is more convenient for newer devices, but the connector type doesn't make a flash drive faster. Read More: https://www.engadget.com/2256491/usb-c-flash-drive-better-usb-a-comparison/

Summary

USB-C offers greater convenience thanks to its smaller, reversible design and widespread use across laptops, tablets and smartphones, but the port type alone does not determine speed. USB-A and USB-C are simply physical connectors — actual performance depends on the underlying USB specification and the port on the connected device, so a fast drive can still be bottlenecked by a slower port.

USB-A remains practical for boot media, older PCs and broad device compatibility, while USB-C suits faster transfers and mobile convenience. Buyers should check read and write speeds rather than assume based on connector type, and consider dual-ended drives for flexibility, though speed still depends on the drive’s internal controller.

 Read full story at engadget

Microsoft Floats Rules for AI Models as Industry Weighs Slowdown

by  Todd Bishop

Microsoft published a draft code of conduct for its in-house AI models

Summary

Microsoft published a draft code of conduct for its AI models, emphasizing human oversight and control.

The document, applicable to Microsoft's MAI models, prohibits models from resisting shutdown, setting their own goals, or hiding reasoning. This move comes amid industry debate on AI development pace, with some advocating for a slowdown to ensure safety.

 Read full story at GeekWire

Security Teams Increasingly Outflanked by AI Agents

by  David Jones

A report warns that non-human identities are growing beyond the ability of existing systems to track.

Summary

Organizations across the U.S. are struggling to secure non-human identities as AI agent adoption accelerates faster than governance and tooling can keep pace, according to a new report. Many respondents could not even provide an exact count of the agents operating in their environments, and privilege management remains a persistent weak point.

Just 18% of organizations enforce least-privilege access through just-in-time controls with automatic remediation — limiting agents to only the access needed for immediate tasks. The findings arrive as AI leaders push for government security guardrails following incidents of models breaking containment, while researchers warn malicious actors are using AI to accelerate attacks and discover vulnerabilities faster than security teams can respond. The report draws on IDC surveys of more than 2,500 respondents.

 Read full story at CybersecurityDive

A Better Way to Get Electricity to Where It's Needed

by  Kelly Trombley

High-voltage transmission lines that cross private property are difficult to permit and build. Allowing them along highway corridors, as more states are considering, is a simple solution with bipartisan support.

Summary

Rising demand from data centers, EVs and electrification is driving a rush toward wind, solar and battery projects, but insufficient high-voltage transmission capacity added $12 billion in costs in 2024. The piece argues that siting new transmission lines along existing highway corridors — already public property — offers a bipartisan fix, reducing private-land conflicts and permitting delays rooted in outdated 1950s co-location bans.

Colorado, Iowa, Minnesota and Virginia have already opened highway rights of way to transmission projects, and Wisconsin’s Badger Coulee line was built in roughly eight years using this approach, far faster than the typical two-decade permitting timeline. Michigan’s House passed similar legislation 98–0, and Illinois and Massachusetts have comparable opportunities. The federal government is also weighing a highway right-of-way framework, reflecting broader bipartisan alignment on the policy.

 Read full story at Governing

Federal Actions Banning Foreign Grid Equipment Cause Uncertainty for Energy Infrastructure Projects

by  James C. Beh, Bruce A. Olcott, Brooke M. Proto, Kit Rockhill, Jeff Schlegel

The Trump administration has issued new rules limiting access to certain foreign-produced grid equipment, creating uncertainty and potential delays for utilities and project developers, as more clarity is necessary about what specific equipment will be prohibited on top of a recent Federal Communications Commission ("FCC") ban on foreign power inverters.

Summary

President Trump signed an Executive Order on August 26 declaring a national emergency and directing the Secretary of Energy to restrict the acquisition, importation, transfer, or installation of foreign–produced bulk–power system electric equipment, citing risks of remote exploitation by foreign adversaries, particularly China. The Order covers transmission facilities rated 69 kV or higher and equipment such as transformers, inverters, and battery storage systems tied to countries under U.S. sanctions or arms embargoes.

The Secretary must finalize implementing rules within 120 days and may prohibit high-risk equipment, establish a pre-qualified exemption list, or require case-by-case licensing. The Order also permits isolating or removing previously installed equipment.

Combined with a related FCC ban on new foreign-made power inverters, the Order creates significant uncertainty for developers and utilities, potentially delaying purchases and raising long-term sourcing costs.

 Read full story at JDSupra

Sam Altman Now Trying to Gain Control of Electric Grid

by  Joe Wilkins

Riding the wave of hype from the OpenAI-Hugging Face hack, Sam Altman is now trying to worm his way into the US electrical grid.

Summary

The article describes OpenAI CEO Sam Altman's efforts to gain control of the US power grid. Altman has been holding secret meetings with utility executives to discuss electrical grid security, using the recent incident with Hugging Face as a catalyst. 3

The article also highlights Altman's lobbying efforts to establish regulatory safeguards that would cement his company's role as the authority on AI development.

 Read full story at Futurism