LinkedIn Leads Social Platforms in AI-Generated Long-Form Posts

A Pangram study of over one million posts finds 41% of LinkedIn's long-form content is AI-written, far outpacing Reddit, Substack, and X.

LinkedIn accounts for nearly two-thirds of all AI-generated long-form content detected across five major social platforms, even though it made up only a third of total posts scanned, according to a new study by AI-detection firm Pangram [1].

The finding matters for developers and businesses that rely on LinkedIn for professional outreach, recruiting, or thought leadership: the signal-to-noise ratio on the platform may be significantly worse than users assume [1].

What the data shows

Pangram’s Chrome extension scanned more than one million posts across five platforms between April and June 2026 [1]. Overall, one in four social media posts longer than 250 words was flagged as AI-generated [1].

LinkedIn was the worst offender, with 41 percent of its long-form posts flagged as AI-written [1]. On X (formerly Twitter), close to half of long-form articles were identified as AI-generated or AI-assisted [1]. Substack had the lowest long-form AI rate at roughly 10 percent [1].

Reddit showed a split picture: replies were 98 percent human-written, but standalone posts contained AI text far more often [1].

Detection methodology and caveats

Pangram says its Pangram 3 detection model carries a false positive rate — instances where human writing is incorrectly flagged as AI — of just 0.01 percent [1]. However, the company acknowledges the model is likely better at identifying human-written content than AI-generated content, meaning the true AI rate across platforms could be even higher than the study reports [1].

The study does not make any claims about the quality of the content it flagged [1].

LinkedIn appears aware of the problem: the platform has already begun taking steps to crack down on AI-generated posts, according to The Decoder’s reporting [1].


Sources

  1. The Decoder — LinkedIn is the undisputed king of long-form AI slop, according to a study spanning five platforms

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

The AI company is launching its own preclinical research programs targeting diseases that traditional pharma won't touch, framing the work as both a mission play and a way to sharpen its science AI tools.

Anthropic is launching its own drug discovery programs focused on neglected diseases — conditions that traditional pharmaceutical and biotech firms have passed on as unprofitable — with work beginning at the preclinical stage [1].

The company says the initiative fits its nonprofit mission and will also give it firsthand experience to build better AI models and tools for the broader industry [1]. The announcement came during an event for Anthropic’s new science AI tool, Claude Science [1].

What Claude Science can already do

The launch event included early demonstrations of AI-assisted research. A researcher at the University of California, San Francisco used Claude Science to detect a viral contamination in minutes — a problem his team had failed to catch for an entire year, according to Anthropic [1]. The company also says Claude analyzed 100 rare genetic diseases in under an hour and identified 32 candidates for computational screening [1].

How much time AI could actually cut from drug development

Novartis chief executive Vas Narasimhan, who spoke at the event, said getting a finished drug candidate from development to regulatory approval currently takes about twelve years [1]. He broke that delay into three buckets: information latency, operational latency, and biological latency [1].

New AI tools could significantly reduce the first two categories, which together account for roughly 40 percent of total development time, potentially bringing timelines down to seven or eight years [1]. Biological latency — the time required for animal testing, cell models, and human clinical trials — is unlikely to shrink much [1].

Narasimhan also sees a path to doubling drug success rates from 8 to 16 percent through better safety predictions and optimized molecular properties, though he noted the effect of improved patient selection remains unclear [1]. The hardest problem, he said, is still determining whether a drug target is biologically the right one for a given disease [1].

To put the stakes in context: the major pharmaceutical companies collectively spend $150 to $200 billion a year on research and development and have produced only 800 to 1,000 drugs over 120 years, according to Narasimhan [1]. Even modest efficiency gains, scaled across that base, could open up previously unreachable drug targets, he said [1].

A crowded field with cautious observers

Anthropic is not alone in pushing AI into drug development. DeepMind chief executive Demis Hassabis co-founded Isomorphic Labs with Alphabet to apply AI directly to drug discovery, and DeepMind’s protein-structure prediction tool AlphaFold remains one of the most prominent examples of AI in biology [1]. AlphaFold co-developer John Jumper recently left for Anthropic [1].

On the clinical side, Google DeepMind introduced an AI Co-Clinician in 2026 built around what it calls triadic care, where AI agents support patients throughout treatment while the physician retains clinical authority [1]. OpenAI launched ChatGPT Health in early 2026, a dedicated section within ChatGPT that lets users connect medical records, Apple Health data, and wellness apps [1].

Independent researchers urge caution, particularly when AI is applied in clinical settings for diagnoses, treatment plans, and direct patient care. Catherine Pope of the University of Oxford described results so far as “a piece removed from the messy, complex, human world of everyday healthcare” [1].


Sources

  1. The Decoder — Anthropic launches its own drug discovery programs to tackle diseases Big Pharma considers unprofitable

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

A new framework from Alibaba researchers decomposes complex tasks iteratively so AI agents fetch only the tools they need, slashing context costs from ~884,000 tokens to ~1,160 per query.

Alibaba researchers have published SkillWeaver, a framework that routes AI agents to the right tools through iterative task decomposition rather than loading an entire tool library into context, reducing token consumption by over 99% compared to brute-force approaches [1].

For developers building enterprise agents that orchestrate dozens or hundreds of tools, the practical stakes are direct: fewer tokens mean lower API costs and faster responses, and the accuracy gains address a failure mode that currently makes large tool libraries nearly unusable [1].

The problem with loading every tool

Modern large language model (LLM) agents are increasingly integrated with massive tool ecosystems — including systems built on the Model Context Protocol (MCP), a standard for connecting agents to external services [1]. When an agent needs to handle a multi-step request like “Download the dataset, transform it, and create visual reports,” exposing the full tool library to the LLM to find the right tools is, as the researchers describe it, “highly inefficient” and “quickly overwhelms context limits” [1].

Most existing frameworks treat tool selection as a one-shot, single-skill problem, which breaks down when real business queries require sequencing multiple tools in a compatible chain [1].

How SkillWeaver works

SkillWeaver structures tool selection across three stages: Decompose, Retrieve, and Compose [1]. First, an LLM breaks a complex user query into atomic sub-tasks, each requiring exactly one tool [1]. An embedding model then searches the tool library to produce a shortlist of candidates for each sub-task [1]. Finally, a planner checks inter-tool compatibility — whether the output of one tool feeds cleanly into the next — and assembles the result as a Directed Acyclic Graph (DAG), allowing independent steps to run in parallel [1].

The central innovation is a feedback mechanism the team calls Skill-Aware Decomposition (SAD) [1]. Rather than decomposing a task once and searching, SAD has the LLM draft an initial plan, runs a preliminary search to surface loosely matching tools, and feeds those results back to the LLM so it can rewrite its decomposition using the actual vocabulary of the available tools [1]. The authors shared the prompt templates in their paper, and the technique can be reproduced with standard libraries such as LangChain, LlamaIndex, or plain Python [1].

Benchmark results

The team evaluated SkillWeaver on CompSkillBench, a custom benchmark of 300 multi-step queries run against a library of 2,209 real-world skills drawn from the public MCP ecosystem across 24 categories including cloud infrastructure, finance, and databases [1].

Without SAD, a 7-billion-parameter Qwen2.5-7B-Instruct model correctly predicted the number of required steps only 51.0% of the time; activating SAD pushed that to 67.7%, and using the larger Qwen-Max model brought it to 92% [1]. On hard tasks requiring four to five tools, SAD improved accuracy by 50% [1].

One counterintuitive finding: a 14-billion-parameter model performed worse than the 7B model in the unguided setup because it tended to over-decompose tasks into unnecessary micro-steps [1]. SAD corrected this by anchoring the larger model to the actual tool vocabulary, suggesting that tool alignment can matter more than raw model size [1].

The token savings were stark. The brute-force baseline — feeding all tool names into a large model’s prompt — consumed an estimated 884,000 tokens per query and still only retrieved the correct tool category 21.1% of the time [1]. SkillWeaver’s targeted approach brought that figure down to roughly 1,160 tokens per query, a 99.9% reduction, while substantially outperforming on accuracy [1]. A ReAct-style agent baseline achieved 0% decomposition accuracy, collapsing multi-step plans into isolated actions [1].

On retrieval infrastructure, the researchers used the open-source all-MiniLM-L6-v2 embedding model with a FAISS index [1]. Indexing all 2,209 skills took 15 seconds, and retrieval adds under 15 milliseconds of latency per query [1]. Swapping in a stronger encoder (BGE-base-en-v1.5) improved accuracy without any fine-tuning [1].

Limitations and what’s not included

The framework currently has no error-recovery mechanism [1]. If a tool call fails mid-chain, the entire execution breaks, and the authors acknowledge that practitioners will need to build their own retry and fallback logic on top of the compose stage [1].

The source code has not yet been released, though the researchers note the system was built entirely from off-the-shelf components and the SAD prompt templates are available in the paper [1].


Sources

  1. VentureBeat — New Alibaba AI framework skips loading every tool, cutting agent token use 99%

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

The AlphaFold co-creator is departing Google DeepMind after nearly nine years to join Anthropic, in a notable talent shift across the AI industry.

John Jumper, who shared the 2024 Nobel Prize in Chemistry for his work on AlphaFold, announced on Friday that he is leaving Google DeepMind to join Anthropic. [1]

Jumper said he spent “nearly 9 years” at DeepMind before making the move. [1] AlphaFold is an artificial intelligence model capable of predicting the three-dimensional structure of proteins from their genetic sequences — work that earned Jumper and DeepMind chief executive Demis Hassabis the Nobel Prize in Chemistry in 2024. [1]

In a post on X, Jumper credited Hassabis directly for his career trajectory, writing that Hassabis “took a real chance letting me lead the AlphaFold team just six months after finishing my PhD.” [1] He added that DeepMind “is a special place,” suggesting the departure is amicable. [1]

Beyond his protein-structure research, Bloomberg reported that Jumper had been a key member of Google’s team working on coding tools, a product area the company has struggled to sell to businesses. [1]

Jumper’s departure is not the only high-profile exit from DeepMind this week. Character AI co-founder Noam Shazeer also announced he is leaving the company, though Shazeer is heading to OpenAI rather than Anthropic. [1]


Sources

  1. TechCrunch — Nobel laureate John Jumper is leaving DeepMind for rival Anthropic

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

A new decentralized framework from Stanford replaces the traditional central-controller model for multi-agent AI systems, cutting per-task costs roughly in half on software engineering benchmarks.

A Stanford research framework called DeLM — short for decentralized language model — challenges one of the foundational assumptions of modern multi-agent AI systems: that a central orchestrator is required to coordinate agent activity. [1] According to the researchers, eliminating that central controller can cut per-task costs by roughly 50% while improving accuracy. [1]

In a conventional centralized multi-agent setup, a main agent breaks work into subtasks, dispatches them to sub-agents, waits for results, then merges and rebroadcasts context before issuing the next round of instructions. [1] The Stanford team argues this architecture scales poorly: as the number of subtasks grows, the central controller becomes a communication and integration bottleneck, and may “dilute, omit, or distort” useful information in the process. [1]

DeLM replaces that hub-and-spoke model with three components: parallel agents, a shared context store, and a task queue. [1] The shared context holds compressed information summaries — called “gists” — that include verified findings, partial findings, and documented failures, along with pointers to detailed evidence agents can retrieve on demand. [1] Agents write results directly into this shared store rather than routing them through a central controller. [1]

The pipeline works in five stages: inputs are broken into work units and queued; agents pull tasks and read shared context in parallel; results are compressed into gists and verified before being shared; a final agent checks whether additional work is needed; and that agent returns the answer once no further steps are required. [1]

A key design feature is what the researchers call “unfolding” — agents see short gists by default but can expand them into fuller summaries or raw evidence when needed. [1] Co-developer Yuzhen Mao explained the tradeoff: “If agents shared full traces, each worker would need to read long command histories, file dumps, failed edits, and intermediate reasoning, turning coordination itself into another long-context bottleneck.” [1] Conversely, sharing only compact summaries risks losing critical details. [1] The opt-in unfolding mechanism is designed to balance cost and accuracy. [1]

The framework also addresses a specific inefficiency in parallel agent runs: when one agent pursues a dead-end reasoning path, that failure is normally invisible to other agents, which may then repeat the same mistake. [1] With DeLM, failed hypotheses are written into shared context so that, as Mao noted, “later agents can read them as constraints, avoid repeated exploration, and redirect their search toward more promising fixes.” [1]

On SWE-bench Verified — a benchmark that evaluates AI systems on real-world software engineering problems — DeLM performed 10.5% better than the strongest baseline and reduced cost per task by roughly 50%. [1] On LongBench-v2 Multi-Doc question answering, which tests long-context reasoning across multiple documents, DeLM achieved the highest accuracy across four model families, including GPT-5.4, Claude Sonnet, Gemini Flash, and DeepSeek-V4-Pro. [1]

The framework is described as particularly suited to software engineering test-time scaling — where models are given additional compute time to improve reasoning — as well as long-context tasks where multiple agents can examine separate evidence clusters while maintaining a shared global view of accumulated findings. [1]

For developers and enterprise teams building multi-agent workflows, the practical implication is that the central-orchestrator pattern, while intuitive, carries measurable costs in both inference spend and coordination latency that a shared-state, decentralized design may avoid. [1]


Sources

  1. VentureBeat — Stanford's DeLM cuts multi-agent task costs 50% — without a central orchestrator

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

A new Google paper reframes hallucinations as 'confident errors' and introduces a metacognitive technique that lets models hedge answers rather than guess or refuse.

Google researchers have published a paper introducing a concept called “faithful uncertainty,” a technique designed to reduce hallucinations in large language models (LLMs) without stripping away their usefulness — a tradeoff that has long frustrated enterprise AI developers. [1]

The core problem the paper addresses is what the authors call the “utility tax.” Current hallucination-mitigation strategies typically force a model into an “answer-or-abstain” binary: either respond confidently or say nothing. [1] The researchers demonstrate that reducing an underlying 25% error rate to a strict 5% target requires discarding 52% of the model’s correct answers. [1] In practice, developers faced with that tradeoff tend to prioritize coverage, leaving models that continue to produce confident hallucinations. [1]

To escape that bind, the paper reframes hallucinations not as any factual error, but specifically as “confident errors” — incorrect information delivered authoritatively without appropriate qualification. [1] Under this definition, a model that says “I am not completely sure, but I think…” before giving a wrong answer is not hallucinating; it is offering a hypothesis. [1] A model that states the same wrong answer as established fact is. [1]

“Faithful uncertainty” is the mechanism the researchers propose to operationalize that distinction. [1] It requires aligning a model’s linguistic uncertainty — the words it uses to express doubt — with its intrinsic uncertainty, meaning its actual internal statistical confidence in a given answer. [1] This ensures hedging language appears only when the model’s internal state genuinely reflects low-probability or conflicting information, rather than as a blanket disclaimer on every response. [1]

Gal Yona, a research scientist at Google and co-author of the paper, told VentureBeat that the two main paths to improving factuality — training models on more data and building faithful uncertainty — are complementary rather than competing. [1] Knowledge expansion pushes the boundary of what a model knows outward, while faithful uncertainty communicates honestly wherever that boundary currently sits. [1]

The technique has particular relevance for agentic AI systems, where models orchestrate external tools such as search application programming interfaces (APIs). [1] Without metacognitive awareness of their own confidence levels, agents must rely on external, static heuristics to decide when to retrieve information. [1] Yona described today’s approaches as “static and brittle,” noting that a model might search for something it already knows — wasting latency and cost — or answer from memory when it should have searched, producing a plausible but wrong output. [1] Faithful uncertainty would allow an agent to trigger a search tool only when its internal confidence is genuinely low. [1]

The paper also identifies a significant implementation challenge it calls the “bootstrapping paradox.” Teaching a model the syntax of uncertainty through supervised fine-tuning (SFT) requires training data, but the correct expression of uncertainty depends on what a specific model knows at a specific point in training — a moving target. [1] “If you train on a label that says ‘I don’t know X’ but the model actually does know X, you’ve taught it to hallucinate uncertainty,” Yona said. [1]

For enterprise developers who want to experiment without retraining models, Yona pointed to prompt engineering as the lowest-friction starting point, and noted that an open-source framework called MetaFaith — which he previously co-authored — can be applied to off-the-shelf models. [1] However, he cautioned that “there is still substantial headroom that prompting alone doesn’t solve,” and that the industry will eventually need reinforcement learning (RL) to embed metacognition more deeply into model training. [1]

Evaluating whether a model has genuinely internalized uncertainty awareness, rather than learned to mimic its surface style, remains an open research problem. [1] “A model might learn to mimic the style of uncertainty without truly sensing its internal state,” Yona said, adding that developing evaluation frameworks capable of telling the difference is “one of the most important open problems in this space.” [1]


Sources

  1. VentureBeat — Google researchers introduce 'faithful uncertainty,' allowing LLMs to offer best guesses instead of hallucinations

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.