SecurityBrief UK - Technology news for CISOs & cybersecurity decision-makers
United Kingdom

AI Safety Stories - Page 13

UK Stories

AI Safety stories

UK government launches GBP £500m Sovereign AI Unit
Semiconductors

UK government launches GBP £500m Sovereign AI Unit

The £500 million fund is meant to help British AI start-ups scale, as ministers seek growth and greater control over core technology.

By Shannon Williams • 4 min read • Last month
UK backs Cosine with GPU hours for sovereign AI push
Hyperscale

UK backs Cosine with GPU hours for sovereign AI push

The grant lets the London startup train an air-gapped coding model on UK infrastructure, bolstering supply for defence and other sensitive sectors.

By Mark Tarre • 4 min read • Last month
Anthropic's Mythos AI sparks UK bank cyber stability alarm
Firewalls

Anthropic's Mythos AI sparks UK bank cyber stability alarm

UK regulators are racing to assess whether Anthropic’s Mythos model could speed up attacks on banks and unsettle financial stability.

By Karen Joy Bacudo • 5 min read • Last month

Global Stories

AI Safety stories

Mythos changes everything: Is your AI agent security ready?
Firewalls

Mythos changes everything: Is your AI agent security ready?

Enterprises face a new security gap as AI agents spread without oversight, with one preview model finding attack paths in hours rather than days.

By Dev Rishi • 5 min read • Last month
OpenAI launches GPT-Rosalind for life sciences research
Data Analytics

OpenAI launches GPT-Rosalind for life sciences research

Research teams could see faster target discovery as OpenAI opens GPT-Rosalind to qualified US customers for biology and drug discovery work.

By Mark Tarre • 4 min read • Last month
OpenAI launches Trusted Access for Cyber with major names
Firewalls

OpenAI launches Trusted Access for Cyber with major names

Banks and security firms will test how advanced AI cyber tools can aid defence without widening the risk of offensive misuse.

By Mark Tarre • 4 min read • Last month
Anthropic launches Claude Opus 4.7 with stronger coding
DevOps

Anthropic launches Claude Opus 4.7 with stronger coding

The update promises better software engineering and longer task handling for users, while keeping Claude Opus 4.7 at the same price.

By Mark Tarre • 4 min read • Last month
Testlio launches AI chatbot testing service amid scrutiny
Data Protection

Testlio launches AI chatbot testing service amid scrutiny

Brands using customer-facing chatbots face fresh pressure to prove safety and accuracy as Testlio rolls out human-led checks for live-use failures.

By Sofiah Nichole Salivio • 3 min read • Last month
Mastercard & Lobster.cash team up for AI agent payments
Crypto

Mastercard & Lobster.cash team up for AI agent payments

OpenClaw users will be able to let AI agents pay with existing cards as Mastercard's controls add limits, authentication and audit trails.

By Karen Joy Bacudo • 4 min read • Last month
Trust fears stall AI projects at medium & large firms
Data Protection

Trust fears stall AI projects at medium & large firms

Trust concerns are pausing nearly half of planned AI spending at medium and large firms, with explainability now outweighing regulatory uncertainty.

By Joseph Gabriel Lagonsin • 4 min read • Last month
Rubrik survey warns of AI agents outpacing security
Data Protection

Rubrik survey warns of AI agents outpacing security

Most firms expect autonomous tools to outstrip guardrails within a year, leaving agent actions hard to see, control and roll back.

By Joseph Gabriel Lagonsin • 4 min read • Last month
The autonomous SOC: A dangerous illusion as firms shift to human-led AI security
Data Protection

The autonomous SOC: A dangerous illusion as firms shift to human-led AI security

Businesses risk disruption if they hand security decisions to AI, as experts argue human oversight is needed to keep responses in context.

By Moe Ibrahim • 4 min read • Last month
Emergent launches Wingman autonomous AI agent for work
Software-as-a-Service

Emergent launches Wingman autonomous AI agent for work

By handling emails, calendars and routine requests in the background, the tool aims to cut admin for businesses wary of autonomous AI risks.

By Sean Mitchell • 3 min read • Last month

Stories from Other Regions

AI Safety stories

Carbery CogniMed, Love Better win UCC IGNITE awards
EduTech

Carbery CogniMed, Love Better win UCC IGNITE awards

The showcase highlighted early-stage ventures tackling clinical delays, relationship support and school safety as finalists pitched to investors and local firms.

By Joseph Gabriel Lagonsin • 4 min read • Last month
Areto founder Jacqueline Comer wins global AI safety award
Email Security

Areto founder Jacqueline Comer wins global AI safety award

The award lifts Areto’s profile as it expands software that has blocked more than 229,000 fraud attempts and illegal streams in a year.

By Joseph Gabriel Lagonsin • 5 min read • Last month
Autonomize launches healthcare AI platform version 3
Digital Transformation

Autonomize launches healthcare AI platform version 3

Healthcare groups under pressure to cut admin delays are being offered a platform already running at three of the five biggest U.S. health enterprises.

By Mark Tarre • 4 min read • Last month
OpenAI broadens AI cyber tools as arms race heats up
Data Protection

OpenAI broadens AI cyber tools as arms race heats up

Ransomware pressure on US firms is intensifying debate over whether broader AI hacking tools will help defenders or aid criminals.

By Sean Mitchell • 4 min read • Last month