GoKawiil
Tech News
A technical essay examines why using large language models directly as classifiers is frustrating despite decent performance, citing problems with calibration, inability to reliably use structured data or context, and limited interpretability of prompt-following behavior. The author suggests reframing the role of LLM outputs entirely rather than treating them as final classification decisions.
Security firm Strix ran an unauthenticated scan against Baseten's public infrastructure and within 25 minutes uncovered a live GitHub personal access token with admin rights to Baseten's core product repo, GitOps repo, and Homebrew tap. The token, tied to a container image built in March 2023, had gone unnoticed for over three years and still granted read/write access to private customer repositories when discovered in July 2026. Baseten's security team confirmed the issue as critical and rotated the token within a day of disclosure.
A growing number of employers are setting internal limits on how many AI tokens their staff can consume, aiming to keep generative AI spending in check. However, implementing these caps has proven more complicated than expected, as companies struggle to balance cost control with employee productivity and access to useful AI tools.
UK-based AI consultant Grant de Swardt found his Claude Max subscription burning through tokens even when he wasn't using the service. Anthropic investigated and told him a compromised session key had been used to generate unauthorized OAuth tokens, allowing a third party to consume his allotted usage. The company suspended his account, wiped his active sessions, and issued a partial refund but could not explain how the credentials were obtained.
A hobbyist project lets users hover over tokens generated by a transformer language model to see which earlier tokens most influenced each output, by combining attention weights and value vector magnitudes across all heads and layers into a single opacity value. The demo highlights cases like verbatim copying of addresses or dates, showing the source text lighting up strongly when reproduced.
Hunter Biden is preparing to launch a cryptocurrency token called LAPTOP, following an extended run of podcast and media appearances that raised his public profile. The move mirrors the memecoin ventures previously pursued by Donald Trump and his family.
A model named GPT-6 Astra has appeared on OpenRouter, the multi-provider inference marketplace that routes requests across different hosts using modes like Balanced, Nitro, and Exacto. OpenRouter's listing tracks pricing, throughput, latency, uptime, and app usage for the model rather than confirming details about its origin or capabilities.
A technical essay argues that describing large language models as mere next-token predictors is outdated once reinforcement learning with verifiable rewards (RLVR) enters the picture. Unlike pre-training, which only reinforces sequences already present in training data, RLVR lets models generate novel token sequences and learn from evaluating their outcomes. The piece walks through pseudocode contrasting the two training loops to show how post-training changes what the model is actually optimizing for.
New industry data shows token usage for AI models has surged more than 25-fold over the past year, doubling in just the last month, even as per-token costs for high-intelligence models continue to fall. Mid-tier models from providers like Google and Meta are now delivering roughly 90% of the capability of pricier flagship systems such as Anthropic's Claude Opus at around one-sixth the cost, intensifying competition on what analysts call the 'Pareto Frontier' of price versus intelligence.
Anthropic launched Claude Fable 5.1 across its API, cloud platforms and desktop app, alongside Mythos 5.1, a less-restricted version limited to vetted cybersecurity and life-sciences organizations. The company says Fable 5.1 handles multistep coding and scientific workflows more efficiently, using fewer tokens, and posted large gains on internal benchmarks like Terminal-Bench 4.0 and Terminal-Bench-Science 0.1 versus its predecessor.
Bill Gates published an essay urging governments to impose enforceable rules on AI, including a tax on AI token usage to fund support for displaced workers and protected 'Human Reserved' jobs like caregiving. He said tech executives privately worry about AI's disruptive pace but avoid saying so publicly, breaking from the industry's optimistic public stance.
Following the guilty plea of hacker Connor Moucka, who used stolen but valid credentials to breach over 165 Snowflake customer accounts and steal billions of records including AT&T call logs, Snowflake is phasing out password authentication for non-human service accounts. By October 2026, all legacy service users will be forced onto the SERVICE account type, which cannot authenticate with a password at all.