AI Safety stories
Bias in AI is becoming a business risk as researchers warn English-language models still miss cultural nuance in non-Western markets.
AI rollouts are pushing SRE and platform teams to monitor models, manage incidents and fill governance gaps in production.
Australian firms can speed month-end processing from days to an hour by running Kiraa AI on Macs instead of costly cloud servers.
The beta tool lets Claude users trade prediction markets and pay other agents without leaving their coding session, though security risks remain.
The rollout will let law firms and other clients run AI inside governed repositories without weakening confidentiality or audit controls.
Security gaps are slowing business adoption of AI agents, as most technology leaders warn that governance and access controls lag behind their autonomy.
It is aimed at firms that want AI agents to update records and trigger workflows without losing oversight or audit trails.
Buyers of AI tools now have a benchmark to judge testing providers, as CREST's new standard targets gaps in assurance and due diligence.
Governed agents could speed urgent fixes in defence systems while keeping air-gapped code changes auditable, approved and traceable.
Rising disclosure volumes are forcing security teams to predict which flaws attackers will exploit as FIRST broadens its vulnerability conference in Luxembourg.
The guidance could help firms curb AI costs, protect sensitive data and reduce the risk of errors spreading through agent chains.
High-risk AI obligations have been pushed back, but transparency, GPAI and banned-practice rules remain in force for many firms.
More deals and faster lender integrations are expected as Lendi shifts its mortgage platform towards agentic AI for 1,300 brokers.
Governance gaps are slowing enterprise AI adoption, with 79% of Australian DevSecOps professionals saying compliance is getting harder.
Australian enterprises risk extra compliance burdens as AI-made observability tools shift governance obligations onto in-house teams.
Rising scam losses in Australia are pushing businesses to verify texts, images and voice notes before paying or acting on urgent requests.
Targeted fraud is inflicting bigger losses on small firms as AI helps criminals forge convincing invoices and impersonations in Australia.
Shared code channels could keep AI-assisted development visible to whole teams, with human sign-off still needed before production releases.
Nearly 60% of big UK retailers reported AI security incidents last year as rapid adoption outpaced governance and risk checks.
Existing cameras at a remote Côte d'Ivoire healthcare project cut PPE breaches by 41.7% and danger-zone violations by 46.3% in three months.