OpenAI stories
AWS customers will gain limited-preview access to OpenAI models and Codex inside Bedrock, easing enterprise AI deployment and governance.
The new tools could let merchants sell inside AI apps and bill for token use in real time, while tightening fraud checks.
The London startup will use the cash to expand in the US as its AI matching tool gains traction with engineers and employers.
The listing gives regulated AWS customers a faster route to compliant Kubernetes components, avoiding custom hardening and patching work.
ChatGPT users can now buy a discounted two-pack of hardware keys designed to block phishing and protect sensitive accounts.
Merchants may need to adapt product data as AI assistants increasingly shape online shopping and determine which items appear first.
Teams can now switch between ChatGPT, Claude, Gemini and Grok in one workspace as the Boston-based platform adds image and file tools.
Enterprise AI teams can now run multimodal agents across Vultr’s cloud, with NVIDIA’s open model available on GPU clusters or serverless inference.
Trust is emerging as a selling point for finance software as Sage warns that opaque AI can leave CFOs answerable for costly errors.
The release aims to curb a growing security risk as enterprises let autonomous agents into internal apps with broad human-style access.
The open-source spec aims to let teams automate coding work through ticket-driven agent workflows, while reducing context-switching for engineers.
Shoppers can now see cashback offers in real time inside ChatGPT chats, as Minty targets AI-led buying decisions and retail media budgets.
Australian businesses risk vanishing from search as AI tools replace blue links with one answer, cutting clicks and reshaping discovery.
Australian sole traders are using AI to cut admin time and boost output, with daily or weekly use now at 41 per cent, Hnry found.
Incorrect AI responses are already steering customers away, with Atlas finding factual errors in most brand profiles across major platforms.
More than half of early Australian users were trying the image tool for the first time, as prompts skewed towards portraits, anime and simple fixes.
Most Australian security teams lack confidence their controls can spot a compromised AI system, even as firms push assistants beyond pilots.
AI tools now favour recent, credible coverage over paid media, leaving B2B tech firms with a growing visibility gap in search results.
Only 21.1% of workers have had training, leaving many to rely on generative AI at work while still worrying about errors and poor output.
Current frontier models still fall short of stand-alone cyber defence, with the top performer spotting only 46% of attack evidence in Simbian’s test.
