
Multiverse Computing has released ProvenanceGuard, a post-generation verification layer designed specifically for Model Context Protocol (MCP) agents. Unlike traditional verifiers that pool evidence and check for factual support, ProvenanceGuard preserves source identity to detect cross-source conflation—where a true fact is incorrectly attributed to the wrong data source. In evaluations on medical agent traces, the system correctly identified 138 of 139 claims with wrong attributions, outperforming source-blind baselines. The tool integrates with repair loops to block or revise answers based on strict provenance checks, addressing a critical gap in multi-tool AI reliability.
Read original
© WIRED AIAnthropic’s Claude identified a novel reverse transcriptase system in jumbo phages that resembles CRISPR, but the scientific community remains skeptical. While the speed of discovery is impressive, experts note the finding lacks wet-lab validation and may simply be pattern recognition on known data. The real story isn't a new gene-editing tool, but the opaque nature of how an AI model sifts through genomic databases to propose hypotheses that humans must still verify.
© MIT News AIA new MIT study dismantles the alarmist narrative that widespread adoption of a single AI algorithm inevitably leads to systemic exclusion. By modeling hiring scenarios, researchers prove that while monoculture reduces individual discovery, it can actually increase candidate bargaining power and overall hiring volume. The real risk is informational stagnation, which the paper suggests can be mitigated through ensemble methods or injected randomness. This shifts the debate from moral panic to technical optimization of algorithmic diversity.