494 summarised stories in Safety & Ethics, each linking back to the original source. Browse all topics →
OpenAI 1 year ago 16
Anthropic released a system card documenting the safety evaluation and testing conducted before launching its deep research tool. The evaluation included external red teaming exercises and assessments using the company's Preparedness Framework to identify frontier risks. The document details the specific mitigations built into the system to address identified risk areas.
The daily briefing
TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.
TLDRocket
7 August 2026
The infrastructure layer of AI is consolidating around a new constraint: not chips, but the wires connecting them. Lumilens emerged from stealth with a $5.51 billion valuation and $700 million in Series C funding after securing multi-billion-dollar commitments from hyperscalers, a striking validation that optical interconnects—not GPUs—are now the bottleneck in building trillion-parameter clusters. Meanwhile, Nscale is preparing a US IPO on the back of a $51 billion contract backlog and quarterly revenue that jumped from $37 million to over $100 million in six months, betting that vertical integration across power, data centres, GPUs, and software will command a premium from public markets. Both companies are surfing the same wave: the shift from GPU scarcity to connectivity scarcity, and the realization that hyperscalers prefer owning infrastructure end-to-end rather than renting from cloud providers.
Read the full briefing →