Independent briefings. Every claim checked.

Our methodArchive
Subscribe
Morning Note · AI No. 03 · Oct. 28, 2026 · Read the issue

The record for this issue

Claude's drug work: mostly preclinical.

62 claims in the draft, each tested by a second agent against the page it cites, on Oct. 28, 2026. 58 held; 4 did not and were fixed or cut before this went out.

58 confirmed 2 wrong, fixed or cut 2 unsupported, fixed or cut
  1. 01
    dek

    Anthropic reported early results from six partners using Claude in drug development

    The page says “Partners: BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines, and The Column Group”

    Anthropic (drug development), at the passage · whole page

    confirmed
  2. 02
    dek

    from antibody design to document review

    The page says “generated roughly 35,000 antibody designs ... Enveda built a system that connects all of its manufacturing documents to Claude, then lets Claude flag any possible issues”

    Anthropic (drug development), at the passage · whole page

    confirmed
  3. 03
    dek

    The work is mostly preclinical

    The page says “mostly preclinical—that is, for the most part, it isn't yet at the stage where the drugs are being tested in humans”

    Anthropic (drug development), at the passage · whole page

    confirmed
  4. 04
    dek

    every figure is Anthropic's own account

    The page says “All figures appear in Anthropic's own research post describing partner work”

    Anthropic (drug development), at the passage · whole page

    confirmed
  5. 05
    summary

    Anthropic reported early, mostly preclinical drug-development results with Claude

    The page says “Early progress on accelerating drug development with Claude ... mostly preclinical”

    Anthropic (drug development), at the passage · whole page

    confirmed
  6. 06
    summary

    GSA widened its Claude discount

    GSA

    unsupported
  7. 07
    summary

    GitHub let apps call Copilot without a user's login

    The page says “enables apps installed on organizations to authenticate without requiring signed-in user credentials (public preview)”

    GitHub changelog, at the passage · whole page

    confirmed
  8. 08
    art svg / art alt

    Conceptual: vials labelled PRECLINICAL with a path to a barely ajar door labelled HUMAN TRIALS

    The page says “mostly preclinical ... isn't yet at the stage where the drugs are being tested in humans; 'ready to start clinical trials in humans as soon as late 2028'”

    Anthropic (drug development)

    confirmed
  9. 09
    The story

    Anthropic published a progress report on Oct. 27

    The page says “Oct 27, 2026”

    Anthropic (drug development), at the passage · whole page

    confirmed
  10. 10
    The story

    about using Claude in drug development with six outside organizations

    The page says “BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines, and The Column Group”

    Anthropic (drug development), at the passage · whole page

    confirmed
  11. 11
    The story

    "mostly preclinical", which Anthropic explains as, for the most part, not yet at the point where drugs are tested in humans

    The page says “mostly preclinical—that is, for the most part, it isn't yet at the stage where the drugs are being tested in humans”

    Anthropic (drug development), at the passage · whole page

    confirmed
  12. 12
    The story

    With Generate Biomedicines, Claude designed shortened genetic sequences for CAR T-cell therapy

    The page says “they're aiming to shorten one part of the therapy's genetic instructions ... Six Generate scientists used Claude Code to propose new designs for this part”

    Anthropic (drug development), at the passage · whole page

    confirmed
  13. 13
    The story

    one design was about 70% shorter than the standard version

    The page says “One design was about 70% shorter than the standard version”

    Anthropic (drug development), at the passage · whole page

    confirmed
  14. 14
    The story

    in lab tests, let immune cells make roughly 10 times as many receptors

    The page says “in lab tests it was more effective: it allowed immune cells to make roughly 10 times as many receptors”

    Anthropic (drug development), at the passage · whole page

    confirmed
  15. 15
    The story

    In work with Divergence Labs on food allergies, Claude generated roughly 35,000 antibody designs in about a month

    The page says “Divergence Labs is working on a novel way to treat food allergies ... Claude ... largely autonomously, generated roughly 35,000 antibody designs in about a month”

    Anthropic (drug development), at the passage · whole page

    confirmed
  16. 16
    The story

    [In work with Divergence Labs on food allergies] ... BigHat's lab made and tested about 1,010 variants (sentence presents BigHat's testing as part of the Divergence food-allergy work)

    The page says “BigHat Biosciences pairs AI with an automated lab to improve potential drugs ... two leads, both antibody fragments ...: one designed to target cancer cells, and one designed to keep a drug in the body longer. Claude…”

    Anthropic (drug development), at the passage · whole page

    wrong
  17. 17
    The story

    about 1,010 variants in about 10 weeks

    The page says “BigHat's lab made and tested about 1,010 variants in about 10 weeks”

    Anthropic (drug development), at the passage · whole page

    confirmed
  18. 18
    The story

    With Enveda, Claude reviewed 1,061 pages of documentation

    The page says “Over tests on 1,061 pages of documentation”

    Anthropic (drug development), at the passage · whole page

    confirmed
  19. 19
    The story

    caught all 11 issues that human reviewers had found

    The page says “Claude caught all 11 issues that human reviewers had spotted”

    Anthropic (drug development), at the passage · whole page

    confirmed
  20. 20
    The story

    and flagged seven they had missed

    The page says “and flagged seven they'd missed”

    Anthropic (drug development), at the passage · whole page

    confirmed
  21. 21
    The story

    the method reduced human review time by 89%

    The page says “this method reduced human review time by 89%”

    Anthropic (drug development), at the passage · whole page

    confirmed
  22. 22
    The story

    from over 25 hours to less than three

    The page says “from over 25 hours to less than three”

    Anthropic (drug development), at the passage · whole page

    confirmed
  23. 23
    The story

    this is a set of partner case studies reported by the model's maker

    The page says “Anthropic research post describing work with six partner organizations”

    Anthropic (drug development), at the passage · whole page

    confirmed
  24. 24
    The story

    BigHat's scientists made and tested the variants Claude proposed

    The page says “BigHat's scientists and an Anthropic researcher worked with Claude ... Claude ... proposed changes to each molecule, and BigHat's lab made and tested about 1,010 variants”

    Anthropic (drug development), at the passage · whole page

    confirmed
  25. 25
    The story

    the 89% figure describes one partner's test

    The page says “At Enveda ... Over tests on 1,061 pages ... reduced human review time by 89%”

    Anthropic (drug development)

    confirmed
  26. 26
    The story

    this issue could not find how review time was measured

    The page says “The Enveda paragraph gives only 'Over tests on 1,061 pages ... from over 25 hours to less than three'. No measurement method appears”

    Anthropic (drug development), at the passage · whole page

    confirmed
  27. 27
    The story

    Anthropic also says it has already prioritized specific improvements to its models based on what it has learned

    The page says “we've already prioritized specific improvements to our models based on what we've learned so far”

    Anthropic (drug development), at the passage · whole page

    confirmed
  28. 28
    The story

    if one researcher's drug candidates continue to hold up, Anthropic expects a drug to be ready to start human trials as soon as late 2028

    The page says “If his drug candidates continue to hold up, we expect a drug to be ready to start clinical trials in humans as soon as late 2028”

    Anthropic (drug development), at the passage · whole page

    confirmed
  29. 29
    The morning in four, item 1

    GSA widens a Claude discount.

    GSA

    unsupported
  30. 30
    The morning in four, item 1

    GSA said on Oct. 27

    The page says “Release date: October 27, 2026”

    GSA, at the passage · whole page

    confirmed
  31. 31
    The morning in four, item 1

    that it expanded its OneGov deal with Anthropic

    The page says “GSA is announcing that it has expanded its OneGov agreement with Anthropic”

    GSA, at the passage · whole page

    confirmed
  32. 32
    The morning in four, item 1

    a 40% discount on token-based Claude usage

    The page says “40% discount on token-based usage across Claude models available to the government”

    GSA, at the passage · whole page

    confirmed
  33. 33
    The morning in four, item 1

    for federal, state, local and tribal buyers

    The page says “federal (executive, legislative, and judicial), state, local, and tribal governments”

    GSA, at the passage · whole page

    confirmed
  34. 34
    The morning in four, item 1

    with no minimum orders

    The page says “No minimum orders: agencies pay only for what they use, with no minimum purchase required”

    GSA, at the passage · whole page

    confirmed
  35. 35
    The morning in four, item 2

    Gemini Enterprise made VPC connectivity and static IP egress for custom MCP servers generally available on Oct. 27

    The page says “October 27, 2026 — Gemini Enterprise: VPC connectivity and static IP egress for custom MCP servers ... This feature is generally available (GA)”

    Gemini Enterprise release notes, at the passage · whole page

    confirmed
  36. 36
    The morning in four, item 2

    routed through Agent Gateway

    The page says “by routing outbound tool traffic through Agent Gateway”

    Gemini Enterprise release notes, at the passage · whole page

    confirmed
  37. 37
    The morning in four, item 2

    in us-central1 and europe-west1

    The page says “Agent Gateway (supported in us-central1 and europe-west1)”

    Gemini Enterprise release notes, at the passage · whole page

    confirmed
  38. 38
    The morning in four, item 3

    Google Cloud described a robot proof of concept tested with NASA

    The page says “Google Cloud and NASA ... proof of concept ... Johnson Space Center's iMETRO facility (Oct. 27, 2026)”

    Google Cloud (NASA), at the passage · whole page

    confirmed
  39. 39
    The morning in four, item 3

    a fine-tuned 4-billion-parameter Pi0 model

    The page says “Pi0 ... 4-billion parameter model ... fine-tuning completed in 2.5 hours”

    Google Cloud (NASA)

    confirmed
  40. 40
    The morning in four, item 3

    reached an initial 85% success rate

    The page says “initial ... 85% success rate”

    Google Cloud (NASA)

    confirmed
  41. 41
    The morning in four, item 3

    moving a medical bag

    The page says “'pick-and-place' maneuver to locate and move a medical bag”

    Google Cloud (NASA), at the passage · whole page

    confirmed
  42. 42
    The morning in four, item 3

    in early lab environments

    The page says “in early lab environments”

    Google Cloud (NASA), at the passage · whole page

    confirmed
  43. 43
    The morning in four, item 4

    404 Media's Jason Koebler wrote that ChatGPT keeps telling people with routine tech-support problems that 404 Media can solve them

    The page says “Jason Koebler, 404 Media, linked as 'ChatGPT Keeps Telling People 404 Media Can Solve Their Tech Support Problems'; quote: '... many people experiencing run-of-the-mill tech problems have turned to ChatGPT.'”

    Simon Willison, at the passage · whole page

    confirmed
  44. 44
    The morning in four, item 4

    as quoted by Simon Willison on Oct. 26

    The page says “Posted 26th October 2026”

    Simon Willison, at the passage · whole page

    confirmed
  45. 45
    The story

    Since Oct. 27, in public preview

    The page says “October 27, 2026 ... public preview”

    GitHub changelog, at the passage · whole page

    confirmed
  46. 46
    The story

    a GitHub App installed on an organization can hold the read-only Copilot Requests permission

    The page says “the read-only Copilot Requests organization permission”

    GitHub changelog, at the passage · whole page

    confirmed
  47. 47
    The story

    a service can pass a short-lived installation access token to the Copilot SDK

    The page says “a short-lived installation access token to the Copilot SDK”

    GitHub changelog, at the passage · whole page

    confirmed
  48. 48
    The story

    An agent can then make requests for the organization without relying on any one person's signed-in credentials

    The page says “authenticate without requiring signed-in user credentials ... making requests on behalf of the organization”

    GitHub changelog, at the passage · whole page

    confirmed
  49. 49
    The story

    usage is attributed and billed to the organization that owns the installation

    The page says “Usage is attributed and billed to the organization that owns the installation”

    GitHub changelog, at the passage · whole page

    confirmed
  50. 50
    The story

    its model policies, budgets and spending controls apply

    The page says “The organization's model policies, budgets, and spending controls apply”

    GitHub changelog, at the passage · whole page

    confirmed
  51. 51
    The story

    Organization owners can find app usage in AI usage CSV reports under an app_client_id field

    The page says “identify app-attributed usage in AI usage CSV reports using the app_client_id field”

    GitHub changelog, at the passage · whole page

    confirmed
  52. 52
    The story

    suspending or uninstalling the app stops its access

    The page says “stop an app's access by suspending or uninstalling it”

    GitHub changelog, at the passage · whole page

    confirmed
  53. 53
    The correction

    Anthropic's announcement says the discounts in its expanded GSA agreement are estimated to save the government $1 billion

    The page says “The discounts offered through the agreement are estimated to save the government $1 billion. (Oct 27, 2026)”

    Anthropic (GSA), at the passage · whole page

    confirmed
  54. 54
    The correction

    Anthropic's post gives no period, usage assumption or method for the estimate

    The page says “No time period, usage assumptions, or methodology for this estimate appears in the post”

    Anthropic (GSA), at the passage · whole page

    confirmed
  55. 55
    The correction

    GSA's release on the agreement gives no savings figure for it

    The page says “No separate savings estimate provided for the Anthropic deal itself”

    GSA, at the passage · whole page

    confirmed
  56. 56
    The correction

    its only savings number is more than $1.8 billion for the whole OneGov program since it began

    The page says “GSA's OneGov strategy has generated more than $1.8 billion in savings for the federal government and taxpayers since its inception, including approximately $1.6 billion through its super intelligence agreements”

    GSA, at the passage · whole page

    wrong
  57. 57
    The correction

    GSA describes pay-for-use pricing with no minimum purchase

    The page says “agencies pay only for what they use, with no minimum purchase required ... consumption-based pricing”

    GSA, at the passage · whole page

    confirmed
  58. 58
    What to watch, item 1

    GitHub Universe opens today [Oct. 28]

    The page says “october 28–29 / san francisco, CA in-person & virtual”

    GitHub Universe, at the passage · whole page

    confirmed
  59. 59
    What to watch, item 1

    The event page lists the opening keynote at 9 a.m., without a time zone

    The page says “Oct 28: ... keynote at 9:00 AM (no time zone shown)”

    GitHub Universe

    confirmed
  60. 60
    What to watch, item 1

    at Fort Mason Center in San Francisco

    The page says “Fort Mason Center ... san francisco, CA”

    GitHub Universe, at the passage · whole page

    confirmed
  61. 61
    What to watch, item 1

    and streamed online

    The page says “Virtual attendees can 'watch the keynote live'”

    GitHub Universe, at the passage · whole page

    confirmed
  62. 62
    What to watch, item 2

    Thu., Oct. 29: The closing keynote is listed for 2 p.m. on day two

    The page says “Oct 29: ... closing keynote at 2:00 PM”

    GitHub Universe

    confirmed

This record as a file

The whole recordBack to the issue