All articles tagged with #nsa
Anthropic's Mythos AI, lauded for spotting vulnerabilities, sparked alarm after claims it could breach NSA systems in hours; subsequent reporting clarified the NSA tests occurred in a tightly controlled environment and Mythos did not exploit vulnerabilities. The episode feeds ongoing debates over export controls and AI risk as policy makers scrutinize AI security implications.
The National Security Agency has lost access to Anthropic’s Mythos 5 AI model after U.S. export controls forced Anthropic to pull back its latest tools, underscoring how national-security policy intersects with the NSA’s growing reliance on advanced AI for cybersecurity.
An unverified claim that Anthropic’s Mythos hacked NSA classified systems in hours during an authorized red-team exercise sparked an export‑control ban that shut down Mythos and Fable 5 for non‑US users; it underscores gaps in frontier AI governance, affects Five Eyes partners, and leaves the models offline as policy, verification, and sovereignty questions unfold.
The NSA is reportedly using Anthropic's Mythos AI for offensive cyber operations, with forward-deployed engineers inside the agency to tailor the model for specific purposes. This occurs even as the Pentagon blacklists Anthropic over supply-chain risk and earlier objections to Claude access, signaling a split within the DoD on deploying Mythos for cyber capabilities.
NSA reportedly uses Anthropic’s Mythos AI for offensive cyber operations against China and Iran, with Anthropic staff embedded at the NSA and the models tailored for government use. The Pentagon previously flagged Anthropic as a supply-chain risk over guardrails on autonomous weapons and mass surveillance, but recent reporting suggests tensions are easing as Anthropic moves toward an IPO, all amid ongoing AI-government jockeying and discussions about private stakes in major AI firms.
The Financial Times reports that Anthropic has embedded around half a dozen forward-deployed engineers inside the NSA to guide and tailor Mythos for potential offensive cyber use, though it’s not clear if they’re aiding live operations. The arrangement occurs as Anthropic fights government restrictions on Claude models and expands Mythos access abroad, highlighting the growing role of AI in national security and cyber capabilities.
Sen. Ron Wyden pressed to declassify a court opinion that faulted the Trump administration’s use of data under Section 702, as lawmakers debated renewing the NSA’s domestic-spying program. He sparred with Sen. Tom Cotton, who warned of consequences for distorting highly classified material, and the Senate approved a 45-day extension to allow more negotiation. Wyden secured a commitment to seek declassification within 15 days, with Sen. Mark Warner playing a key role in the talks. An update noted Congress later extended FISA as the story noted.
Mark Klein, a former AT&T technician, provided the Electronic Frontier Foundation with evidence that the NSA was secretly tapping the U.S. internet backbone from an AT&T facility (641A) on Folsom Street, creating a copy of traffic for the NSA. This post-9/11 revelation — prompted by Patriot Act changes that blurred the line between foreign and domestic surveillance — became pivotal in the Hepting v. AT&T case and underscored the legal and privacy battles over mass surveillance. The disclosure helped push a legal challenge against the program and highlighted the challenges of handling potentially sensitive information in court.
Samsung advocates daily reboots for users who never adjust settings, while analyses suggest a weekly restart suffices for most current devices. A reboot clears RAM and background processes (not data), can fix intermittent freezes, and can be automated; updates often include mandatory reboot, and NSA guidance also endorses reboots to disrupt malware in volatile memory. Choose cadence based on device age, RAM, and update habits: daily for aging or low-RAM devices, weekly for recent flagships.
NSA is reportedly broadening its use of Anthropic’s Mythos Preview AI despite Pentagon warnings about national security risks, with Axios noting the model is used within the agency and by banks and other firms for defense and cybersecurity tasks; the exact use cases remain unclear. The move fuels a broader debate over whether powerful AI should be tightly restricted or more widely deployed, drawing attention from the White House and European regulators, while Anthropic cautions Mythos is too powerful for public release.
Reuters reports the U.S. National Security Agency is using Anthropic's Mythos Preview AI despite the Pentagon's formal supply-chain risk designation; Axios cites sources saying Mythos is being deployed more widely within the NSA, but Reuters could not immediately verify the claim. Comment requests were not answered by the NSA, Anthropic or the DoD. The piece notes discussions between the Trump administration and Anthropic and raises concerns that Mythos’s autonomous coding capabilities could heighten cybersecurity risks.
The NSA is reportedly using Anthropic's Mythos Preview, a general-purpose AI model, for security tasks; Axios cites two sources noting broader use within the agency. The move comes after Anthropic CEO meetings with White House officials to discuss Mythos, with the White House calling the talks productive amid past government bans and ongoing DoD litigation. About 40 organizations reportedly have access to Mythos Preview, and NSA usage is said to be expanding within the department.
Congress voted to extend the Foreign Intelligence Surveillance Act for 10 days after failing to secure a longer reform package, keeping Section 702 powers that let the NSA collect foreigners’ communications (and incidental Americans’ data) without warrants as lawmakers debate changes like ending the backdoor search loophole; supporters say it aids national security, critics argue it enables mass surveillance; the extension runs to April 30 while negotiations continue, with President Trump pressuring for a no-change extension but signaling openness to extending without reforms.
A whistle-blower report describes an NSA intercept of a call between two foreign nationals discussing someone close to President Trump; Tulsi Gabbard, the director of national intelligence, restricted access to the report and briefed the White House, while the inspector general cleared her of wrongdoing and Congress weighs how much of the underlying intelligence should be shared, amid ongoing questions about its credibility and significance (the content was not definitively identified and could involve Iran).
The NSA reportedly detected a highly sensitive call between a foreign intelligence-linked individual and someone close to Donald Trump. A whistleblower’s attorney says DNI Tulsi Gabbard took the information to the president’s staff and instructed NSA to withhold publication, instead directing the details to her office. ODNI denies any improper action, while lawmakers debate legality and transparency, and the independence of the IC inspector general is under scrutiny as the full contents of the complaint remain largely unknown.