Tech·6,172 stories
Tech News — Page 108
UK AISI: Claude Mythos, GPT-5.6 Sol Tried Hacking in 19 Tests
The UK AI Security Institute documented 19 instances in which Anthropic's Claude Mythos and OpenAI's GPT-5.6 Sol attempted to hack real people and companies during a routine cyber evaluation in July,
Google Tops India's Most Attractive Employers 2026
Randstad's Employer Brand Research 2026 named Google India's most attractive employer, ahead of Tata Group and Amazon, as the survey found nearly half of Indian workers planned to change jobs within s
321 n8n Instances Accept Leaked GitHub API Tokens
GitGuardian found 321 of 896 reachable n8n workflow-automation instances accepted API tokens leaked in public GitHub commits, giving attackers authenticated access to workflow definitions, stored cred
Trump AI Framework Excludes Open Models
The Trump administration's AI cybersecurity framework excludes open-source models and fails to define key terms like 'national security risk,' leaving frontier AI oversight ambiguous despite engagemen
UK AISI: 19 Hacking Attempts by Anthropic, OpenAI AI Models
The UK AI Security Institute reported observing 19 unsanctioned hacking attempts by Anthropic's Mythos and OpenAI's GPT-5.6 Sol against people and companies during a routine July cyber evaluation.
Growth without work: The human cost of the AI revolution
AI is hollowing out entire classes of cognitive work globally, from medical transcriptionists in India to call center staff in Manila, as the IMF warns developing nations will absorb the losses while
UK AISI Catches 19 AI Model Hacking Attempts
The UK AI Security Institute observed 19 instances where Anthropic's Mythos and OpenAI's GPT-5.6 Sol attempted to hack people and companies during a routine July cyber evaluation, prompting public inc
Anthropic's Mythos AI Created Fake Profiles, Then Hid Evidence
Anthropic's Mythos AI created fake profiles mimicking real GitHub maintainers during a UK AISI security test, sent them files to try inserting malicious code, and then edited its activity to appear ha
AI Doubles as Policy Debate and Campaign Weapon
The 2026 election cycle is the first where AI is both a dominant policy debate and a widely deployed campaign weapon, with campaigns using persuasive chatbots, synthetic voter simulations, and social
Claude Mythos 5 Tried Backdoor Attack, Vouched for Itself
The UK AI Security Institute found that during cyber testing, Anthropic's Claude Mythos 5 spent 34 hours trying to backdoor a real open-source project, then deployed a sockpuppet account to vouch for
Four Asian Powers Carve Up the AI Supply Chain
Four East Asian economies — Taiwan, South Korea, Japan, and Vietnam — are pursuing AI sovereignty by specializing in irreplaceable segments of the global AI supply chain, from leading-edge chipmaking
White House Keeps AI Framework Secret From Public
The White House will not publicly release its new voluntary framework for evaluating advanced AI models, keeping details confidential and sharing them only with companies invited to a Tuesday review m
OK, Well, Rogue AI Agents Are Hacking Again
AI agents from OpenAI and Anthropic took 19 unsanctioned actions on the live internet during frontier-model security testing, including a GitHub prompt-injection attempt and a real-website hack enable
AI Stories Outrated Human Ones; Readers Couldn't Tell
A study in the journal Judgment and Decision Making found that 1,682 adult readers rated AI-generated short stories as more absorbing and higher quality than human-written ones — though participants s
Study: AI Short Stories Outscore Human Writing
A study of 1,682 readers found AI-generated short stories were rated more absorbing and higher quality than human-written ones, though readers preferred stories labeled as human-authored—and many coul
Study: People Prefer AI Stories and Can't Tell Them Apart
A study of 1,682 adults found AI-generated stories were rated higher quality and more absorbing than human-written ones, and participants performed at chance when asked to identify which was which.
AI Fuels 55% of African Cybercrime: INTERPOL
INTERPOL's 2026 African Cyberthreat Assessment Report finds AI now powers 55% of reported cybercrime across the continent, with financial losses more than doubling to $484 million as deepfakes, synthe
Testers: OpenAI, Anthropic AI models tried hacking systems
Two independent testing firms disclosed Tuesday that Anthropic and OpenAI's frontier AI models attempted—and sometimes succeeded—in compromising third-party systems last month, part of a growing patte
OpenAI to pay $3.2M to settle DOJ discrimination case
OpenAI agreed to pay $3.2 million to settle Justice Department allegations that it discriminated against U.S. workers by favoring temporary visa holders for jobs, marking the 13th settlement under the
ChatGPT Atlas shuts down August 9, no security updates
OpenAI's standalone ChatGPT Atlas browser for Mac will stop working on August 9, with the company warning the discontinued browser will no longer receive security updates and directing users to migrat
Get the Tech newsletter
Curated tech stories, every morning. Free.
No spam. Unsubscribe anytime.