Generative AI is creating fake and enhanced wildlife images, threatening the credibility of citizen science platforms like iNaturalist and Macaulay Library. Scientists are urging birdwatchers to limit AI use in image editing, fearing contamination of data crucial for species monitoring and conservation.
The Australian Senate has passed legislation to reintroduce human oversight for an aged care algorithm, the Integrated Assessment Tool (IAT), which has been criticized for under-assessing care needs. The bill, co-sponsored by multiple parties, aims to allow assessors to override the algorithm's decisions, addressing concerns raised by elderly Australians, carers, and human rights advocates.
The Australian Senate has passed legislation to reinstate human oversight for an aged care algorithm, the Integrated Assessment Tool (IAT), which has been criticized for under-assessing care needs. The bill, co-sponsored by the Greens, Coalition, and an independent senator, aims to allow assessors to override the algorithm's decisions. The government has proposed some human review in "extenuating circumstances," but critics argue it doesn't go far enough.
JPMorgan Chase and Goldman Sachs have restricted employees outside the US from using advanced AI tools like Anthropic's Claude, citing concerns over governance and national security risks. This move reflects growing global fragmentation in AI access and regulation.
The European Commission is employing its own AI tool, GPT@EC, to assess candidate countries' laws for EU membership alignment, aiming to manage increased workload as Ukraine, Moldova, and others pursue accession. Officials from candidate countries express skepticism, suggesting AI should be limited to translation and basic tasks.
New research on Iron Age remains in Scotland reveals evidence of postmortem brain removal and sharpened limb bones possibly used as tools. The findings, from Loch Borralie, suggest complex funerary practices and highlight the interconnectedness of maritime communities.
MiniCPM5-1B, a 1 billion parameter AI model, is designed for on-device deployment, supporting native tool calling and the Model Context Protocol (MCP). It outperforms comparable open-source models in its size class, offering a 128K token context window and efficient inference capabilities.
