• Dynamic languages may use 1/2 to 1/3 the LLM tokens of static languages on trivial problems.
• More complex tasks show no clear superiority of dynamic or static languages for LLM coding agents.
• Obscure or very dense languages like J score lower or inconsistently in practical evals.
• Language popularity weakly correlates with better correctness and lower cost in LLM tasks.
• Strong claims on best language for LLM coding are unsupported; many evals have flaws and variance.#ProgrammingLanguages #LLM #AIcoding

• Pre-training checkpoint dates estimated by probing daily-fact quizzes on models like GPT-5 and Opus.
• Anthropic Opus 4.7+ models share a knowledge cutoff around December 2025, indicating the same training run.
• OpenAI GPT-5.6 models have a separate checkpoint with a cutoff around February 2026.
• Models self-identify their versions revealing training on past-model outputs and chat data.
• Opus 5's published cutoff is May 2026 but matches earlier models' knowledge cutoff (~January 2026).#AItraining #LanguageModels #ModelCutoffs

• Google Search's new AI summaries produce inaccurate factual info, like false sunset times.
• Link rot, AI intermediaries, and content manipulation threaten internet's archival integrity.
• Disney deleted nearly entire FiveThirtyEight archive after dropping the site in 2025.
• Wikipedia traffic declines as AI scrapes content directly, reducing public donations.
• European countries adopt homegrown digital platforms to ensure technological sovereignty.#DigitalSovereignty #AIImpact #InternetArchive

• Tail-call optimization in C was not common before the 1990s due to calling convention constraints.
• In 2001, Mark Probst added tail-call optimization to GCC with limitations like no indirect call handling.
• Recent tests show modern GCC and Clang compilers successfully perform tail-call optimization as per recent research.
• The technique supports compiling very large numbers of code snippets, beyond previous tooling capacity.
• Gforth has not yet integrated this optimization, while the Python community pioneered its application.#TailCallOptimization #CProgramming #GCC

• Claude models launching in the EU on/after August 2, 2026 will embed AI-generated content marks.
• Marks include imperceptible text watermarks and signed provenance metadata on files.
• Marks apply globally across Claude Platform, API, Claude Code, Cowork, and Tag products.
• Detection tools for Claude’s marks will be provided with future technical documentation.
• Marks signal possible AI processing but don’t confirm original authorship or unchanged content.#AITransparency #ClaudeAI #ContentMarking

• France bans unsolicited telemarketing calls starting August 11, 2026.
• Fines reach €375,000 per illegal call for companies, €75,000 for individuals.
• Consumers can report calls via a government website.
• Exceptions include prior consumer consent and existing contractual relations.
• Morocco fears loss of 40,000-50,000 call center jobs due to French law.#FranceLaw #TelemarketingBan #ConsumerProtection

• Mark Twain invested around $10 million in the Paige Compositor, a complex typesetting machine with over 18,000 parts.
• The Paige Compositor attempted to mimic human compositors mechanically but was impractical and never mass-produced.
• Twain's financial ruin and depression were largely due to the failure of this invention during his prime creative years.
• The Linotype machine surpassed Paige's design by simplifying typesetting, casting entire lines more efficiently.
• Only one Paige machine survives today; Twain's vision of automated typesetting machines was ahead of its time but flawed in execution.#MarkTwain #PaigeCompositor #InnovationFailure

• Ante is a self-contained Rust-based coding agent with a ~15MB binary and no runtime dependencies.
• Runs AI models locally via a pinned llama.cpp engine; supports over 12 providers including OpenAI and Anthropic.
• Publicly benchmarked with 89 tasks at 82.7% accuracy using DeepSeek V4 Flash model.
• Uses ~7x less peak memory, ~9x less CPU, and ~5x less disk I/O than Claude Code in benchmarks.
• Supports interactive, headless, server, and bot gateway modes; open protocol and SDK under Apache 2.0.#AIagents #RustLang #LocalInference

• Trump Media launched Truth API, charging Wall Street firms $60,000-$100,000 monthly for early social media data.
• Trump Media reported a $238 million loss between April and June, over ten times last year’s loss.
• The company's revenue rose 89% to $1.7 million, but losses stem mainly from cryptocurrency ventures.
• Truth API aims to create a durable revenue stream alongside advertising and digital assets.
• Trump Media holds $2 billion in total assets, predominantly in cash, short-term investments, and digital currencies.#TrumpMedia #TruthAPI #WallStreet

• In 1988, the author began a 3.5-year solo around-the-world sail in a 31-foot sailboat.
• The book recounts encounters with pirates, storms, and wars during the voyage.
• It explores the evolving relationship between the solo sailor and the sea.
• The author shares a personal transformation from fearing departure to loving ocean life.
• The book is available online and as a 1.3 MB downloadable ZIP under CareWare, free with conditions.#SoloSailing #AroundTheWorld #SeaAdventure

• England aims to eliminate hepatitis C, having treated 80% of known cases by 2024.
• Hepatitis C-related deaths have fallen 36% in the last decade in England.
• Over 100,000 people diagnosed and treated since 2015 in England.
• Free NHS home-testing kits are available for early hepatitis C detection.
• 30,000+ UK people infected via contaminated blood products from 1970-1991.#HepatitisC #HealthUK #VirusElimination

• World Train Map features 1,249 curated passenger train routes worldwide.
• Covers 120 countries with approximately 317,400 km of railway lines.
• The map focuses on notable routes, not every railway on Earth.
• Interactive platform allows browsing detailed train route information.#WorldTrainMap #RailwayRoutes #TravelTech

• Proprietary LLM APIs return encrypted reasoning (chain-of-thought) blocks replayable across sessions and models.
• Researchers demonstrated recovering hidden reasoning by replaying encrypted traces into weaker, jailbroken models.
• Recovered traces revealed private user data including 704 distinct secrets like API keys and passwords.
• 64 secrets were found exclusively inside reasoning blocks, not visible in chat content.
• Analysis covered models from Anthropic, OpenAI, and Google, showing broad vulnerability of encrypted traces.#AIPrivacy #LLMSecurity #ChainOfThought

• Jay Cooke's 1870 financing for Northern Pacific Railway led to the 1873 Panic and economic depression.
• In 2026, tech companies Microsoft, Oracle, Meta, Alphabet, and Amazon have issued $194 billion in debt for AI infrastructure.
• Google raised $85 billion equity in June 2026, including $10 billion from Berkshire Hathaway, to fund AI build-out.
• DeepMind faces leadership changes and loss of frontier AI status with Google Cloud gaining more compute resources.
• Nvidia partners with major financiers to mobilize $500 billion for AI infrastructure as an investable asset class, risking 25% residual-value financing.#AIInfrastructure #Nvidia #TechFinance

• AT Protocol uses a NoSQL cluster with eventual consistency for scalability.
• Introduces an event log (e.g., Kafka) to sync view servers with the NoSQL cluster.
• Data is organized in user data repositories with collections of JSON document records.
• Every internal service is externalized with public APIs for decentralized backend participation.
• Applications read/write JSON repos, emit events to logs, and update view servers in a circular data flow.#ATProtocol #DistributedSystems #Decentralization

• A Metal capability shim boosts LLM inference in macOS VMs up to 16x faster on M1 Ultra.
• Shim changes Apple GPU capability reports, enabling newer Metal kernels in llama.cpp.
• TinyLlama 1.1B prompt processing was 11.08x faster; token generation 16.36x faster inside VM.
• Tested on macOS 26.6.1, shim achieves near bare-metal performance (up to 99.59%) for some models.
• Shim is experimental, per-process, uses private Metal APIs, and limited to current Apple Silicon and guest versions.#AppleSilicon #GPUAcceleration #MacOSVM

• LFM2.5-2.6B is a 2.69B parameter hybrid model with a 128K context window for on-device deployment.
• It achieves 220 tokens/s inference on Apple M5 Max and 113 tokens/s on AMD Ryzen CPU using under 2.5 GB RAM.
• Trained on 34 trillion tokens with agentic reinforcement learning for tool use and multi-step tasks.
• Available in native, GGUF, ONNX, and MLX formats optimized for various hardware including Apple Silicon.
• Benchmark tests show it outperforms larger models (4x size) in tool use and multi-agent tasks.#LiquidAI #OnDeviceAI #LFM2Model

• Amber Newell of Brookfield, Wisconsin, was pulled over twice at gunpoint due to a Flock camera glitch.
• The camera falsely flagged her car in connection to a homicide investigation on I-94 near Milwaukee.
• Milwaukee police admitted the alert was outdated and should have been removed days earlier.
• Brookfield officers followed protocol given the alert, despite the mistake being a human error.
• Newell's daughter is now fearful of police stops, highlighting the impact of the error.#FlockCameraGlitch #FalsePoliceStop #PoliceTechnology

• Go language focuses on software engineering with strong team-driven development principles.
• Go is a platform with built-in tools: formatter, test framework, dependency management, security.
• Go enforces readability and uniform code format, aiding AI and human code review.
• Go has a strong static type system and comprehensive standard library for secure, reliable code.
• Go guarantees backward compatibility and provides tools for large-scale, maintainable codebases.#GoLang #AIcoding #SoftwareEngineering

• Mojo language reaches stable 1.0 release after initial 2023 launch.
• Mojo 1.0 provides stable foundation for long-term development, minimizing breaking changes.
• New features include Python-style lambda syntax and improved memory safety diagnostics.
• MAX framework updated with easier installation and support for GLM-5.2 and Nemotron-H models.
• Modular plans to open source Mojo compiler and toolchain within 2026.
• Upcoming ModCon event on August 18 will discuss Mojo 1.0 and open source plans.#MojoLang #ModularAI #AIProgramming

• OpenAI's head of ethics left the start-up less than one year after joining.
• The departure occurred before August 10, 2026, as reported by Financial Times.
• No specific reasons for the departure were disclosed in the article.
• The Oxford-based start-up had recently hired the ethics head from OpenAI.
• The exit highlights challenges in retaining top ethics executives in AI start-ups.#OpenAI #AIethics #TechLeadership