Bilingual

Oracle outlines all the ways it could lose the farm it bet on AI
24 days ago
- UN warns about the urgent need for global governance to prevent AI-related catastrophes, as capabilities outpace safety regulations.
- Oracle discloses numerous risk factors that could jeopardize its significant investment in AI technology.
- Anthropic is removing hidden code previously used to detect Chinese competitors, citing planned deactivation of steganography systems.
- Cloudflare will block bots that scrape content from ad-supported web pages, balancing search visibility and AI training data protection.
- Russian actors are impersonating Signal support for phishing attacks; US disrupts Iranian propaganda sites.
- Microsoft's patches fail to address zero-day attacks on on-prem SharePoint, while China enhances smartphone surveillance tools.
- DEF CON's Franklin project recruits hackers to strengthen critical infrastructure, expanding from successful voting village initiatives.
- EQT acquires a majority stake in Acronis at a valuation over $3.5 billion, boosting cybersecurity investments.
- Mikko Hyppönen marks ten years since corporate ransomware began, predicting no end in sight but highlighting stable infosec careers.
- Collabora releases CODE 26.04 with Markdown and AI features, intensifying competition among FOSS office suites.
- GIMP 0.54 is revived via Flatpak for retro-computing enthusiasts, using the original Motif toolkit.
- Bcachefs exits experimental status in a performance-focused release, incorporating more Rust but facing AI-related challenges.
- France's digital sovereignty efforts struggle to reduce Microsoft dependency, as Nextcloud storage adoption contrasts with Office migration hurdles.
- CentOS evolved from a biochemist's hobby into an enterprise default, fueled by community response to Red Hat's Windows endorsement.
- A Netflix developer open-sources Project Headroom, an app designed to significantly reduce AI operational costs.
TinyRetroPad, a 2.5 KB 'Notepad' for Windows
23 days ago
- SoftBank is entering the GPU rental market to support AI training in the US.
- UN warns that global governance is needed to prevent potential AI risks due to rapid technological advancement.
- Oracle acknowledges multiple risk factors in its AI-focused business strategy.
- Claude Sonnet 5.0 avoids controversy by prioritizing safety and cost-effectiveness over cybersecurity.
- Anthropic is discontinuing covert code used to monitor Chinese competitors.
- Russian phishing attacks impersonate Signal support, while US targets Iranian propaganda sites.
- Microsoft's patches fail to fix zero-day attacks on on-prem SharePoint, with China enhancing surveillance tools.
- DEF CON's Franklin project recruits hackers to strengthen critical infrastructure, expanding beyond voting villages.
- EQT acquires a majority stake in Acronis at a valuation over $3.5 billion.
- Ransomware persists after a decade, but cybersecurity offers stable career prospects.
- Collabora releases CODE 26.04 with Markdown support and optional AI features in FOSS office suites.
- GIMP 0.54 is revived via Flatpak for retro-computing enthusiasts using Motif.
- Bcachefs leaves experimental status with performance improvements but faces AI-related challenges.
- France's digital sovereignty efforts struggle against Microsoft's dominance, especially in office software.
- CentOS evolved from a hobby project to a key enterprise OS after community response to Red Hat's shift.
- Netflix engineer open-sources Project Headroom to reduce AI costs significantly.
If You Can Write Acceptance Criteria, You Can Write an AI Routing Policy
21 days ago
- An AI Routing Policy is a documented team decision that assigns specific tasks to the most cost-effective execution path, whether it's a cheaper AI model, a frontier model, deterministic code, human review, or no automation, ensuring accountability and traceability.
- The policy requires defining sufficiency criteria—clear standards for acceptable output, similar to acceptance criteria—for each task class, along with escalation triggers and a designated owner to maintain and enforce the policy.
- Implementing a minimal routing log alongside the policy creates a record of actual usage, costs, and outcomes, enabling task-class attribution and calculation of Return on Tokens (ROT), which helps justify AI spend to finance teams.
- A key insight is that routing decisions should consider whether tasks need probabilistic AI judgment every time or can be handled by deterministic code created once, moving beyond just choosing among AI models.
- The policy makes trade-offs between cost, quality, risk, throughput, autonomy, and adaptability explicit, transforming AI usage from ad-hoc habits into governed, repeatable processes that survive individual departures.
AI 2040: Plan A
16 days ago
- AI companies are racing to develop AI smarter than humans, risking extinction or irreversible power concentration.
- Plan A proposes delaying superintelligence until 2040, making AI research public, and fostering global collaboration to avoid a dangerous race.
- The plan emphasizes research transparency, international agreements, and slow, safe scaling towards superintelligence across multiple companies and countries.
- Current AI development lacks adequate plans for control, potentially leading to human extinction or global dictatorship if unchecked.
- The scenario includes a timeline: US-China deal in 2029, pause at human-expert level AI by 2035, and superintelligence by 2040.
- Policy recommendations stress the need for scenario scrutiny and early negotiations to mitigate risks from AI-driven power concentration.
- AI is disrupting white-collar jobs, with AI companies industrializing training processes and accelerating automation across professions.
- Political and public concern is growing, leading to legislative actions like the AI Transparency Act of 2027, but more ambitious measures are needed.
Ben Bernanke Joins Anthropic Oversight Trust
16 days ago
- Ben Bernanke joins Anthropic's Long-Term Benefit Trust to help ensure AI is developed responsibly for humanity's long-term benefit.
- Bernanke's background includes leading the Federal Reserve during the 2008 financial crisis and winning the 2022 Nobel Prize in Economics.
- The Trust is independent, with trustees holding no equity in Anthropic, and aims to balance commercial success with social good.
- Bernanke will contribute expertise on AI's economic impacts, aiding Anthropic's research and governance decisions.
You Have an AI Working Agreement. Write It Down
13 days ago
- An AI Working Agreement consolidates a team's scattered AI usage rules into a single document, detailing delegation, human oversight, review processes, data boundaries, ownership, and amendment procedures.
- Teams already possess implicit AI rules through habits and templates; codifying them prevents knowledge loss during employee turnover and aids new member onboarding.
- Key components include standing defaults, hard boundaries (data, judgment, relationship), active templates, disclosure norms, concern escalation paths, and review triggers.
- Creating the agreement involves gathering existing decisions, filling gaps, assigning ownership, setting amendment rules, and choosing a central location for accessibility.
- Benefits include reducing single points of failure, enabling consistent decision-making, ensuring compliance with organizational policies, and facilitating team alignment and trust in AI outputs.
You Cannot Defend What You Cannot Inspect
12 days ago
- CIRCUIT is an open-source framework for AI interpretability governance, released to address gaps in current AI governance practices.
- Current AI governance focuses on external controls like risk tiering, registries, and output monitoring, treating models as black boxes without explaining why models make decisions.
- The framework emphasizes interpretability to help during incidents, enabling understanding of model decisions rather than just ownership or deployment details.
- CIRCUIT includes maturity ceilings based on inspectability (e.g., open-weight models have higher ceilings than API or vendor-embedded models) and crosswalks to regulations like EU AI Act and NIST AI RMF.
- It features a 29-question vendor questionnaire to highlight transparency gaps, aiming to create coordinated demand for inspectable models across the security community.
- Released under Apache 2.0, CIRCUIT is open for collective adoption and contribution, with a call for practitioners to test and refine it against real deployments.
Kilo Code has been acquired by Anaconda
10 days ago
- Anaconda acquires Kilo Code to address enterprise AI data and token visibility issues.
- Enterprises lack visibility into AI token usage and data security, risking uncontrolled spending and dependency on single model providers.
- Kilo orchestrates trillions of tokens monthly, highlighting the scale of AI workflows and the need for consolidated management.
- The acquisition aims to eliminate the false trade-off between AI development speed and governance, offering a unified platform.
- Anaconda provides trusted AI foundations, while Kilo adds agentic engineering, enabling any IDE and model with governance.
- Early results show 30-50% token consumption reduction via intelligent routing in the Anaconda Platform.
- The vision promotes open, flexible AI to avoid vendor lock-in, with open source as a key component.
- For builders, it offers tool freedom with governance; for leaders, consolidated visibility and policy enforcement across AI activities.
- The platform covers the full AI development lifecycle, emphasizing speed, capability, and trust for enterprise-scale deployment.
China's Xi Jinping launches new AI alliance, WAICO
8 days ago
- Chinese President Xi Jinping urged international cooperation on AI at the World Artificial Intelligence Conference, opposing single-country dominance and stressing human control.
- China launched the World Artificial Intelligence Cooperation Organisation (WAICO) with 29 founding nations, aiming to shape global AI regulations and promote safe, beneficial technology.
- Analysts suggest Beijing uses WAICO to influence UN AI policies, positioning China as a leader in the Global South amid tech competition with the U.S.
- The U.S.-China rivalry includes a 'Chip War,' with China holding advantages in AI data centers, rare earth minerals, and electricity supply despite semiconductor restrictions.
- WAICO reflects China's strategy to export AI governance norms, countering U.S. influence as it retreats from global tech norm-setting processes.

Hasty Briefsbeta

ai governance

Oracle outlines all the ways it could lose the farm it bet on AI

TinyRetroPad, a 2.5 KB 'Notepad' for Windows

If You Can Write Acceptance Criteria, You Can Write an AI Routing Policy

AI 2040: Plan A

Ben Bernanke Joins Anthropic Oversight Trust

You Have an AI Working Agreement. Write It Down

You Cannot Defend What You Cannot Inspect

Kilo Code has been acquired by Anaconda

China's Xi Jinping launches new AI alliance, WAICO