AI news videos

OpenAI Is Teaching Older Adults To Use ChatGPT

OpenAI announced the Older Adults AI Skills Jam, a free in person learning experience hosted by OpenAI Academy in collaboration with Older Adults Technology Services (OATS) from AARP. OpenAI says it is designed to help older adults use ChatGPT confidently and safely.

Watch on YouTube

Transcript

Sources

Every claim in this video comes from the top ranking coverage of this topic. The claims and where each one came from:

• PER UNIT CITATIONS. HOOK ('OpenAI says it is teaching older adults to use ChatGPT today. Here is the plan.'): article body, 'Today, OpenAI Academy, in collaboration with Older Adults Technology Services (OATS) from AARP is hosting the Older Adults AI Skills Jam: a free, in-person learning experience designed to help older adults use ChatGPT confidently and safely.' Wording change: compressed to a two beat hook and attributed with 'OpenAI says' because the page is OpenAI's own announcement; 'today' is spoken instead of any calendar date, which is accurate because the snapshot was captured on the publication date, 2026-09-16. Second beat is a framing line containing no factual claim, present because the renderer requires two hook beats. || SCENE 1 ('OpenAI says its Academy and OATS are hosting the Older Adults AI Skills Jam today. It is free.'): same sentence as above, 'Today, OpenAI Academy, in collaboration with Older Adults Technology Services (OATS) from AARP is hosting the Older Adults AI Skills Jam: a free, in-person learning experience'. Wording change: the partner is named by its short form OATS in narration because the full legal name plus AARP would form a three item spoken string and would push the scene over the thirty word renderer cap; 'free' is quoted directly from the article, and the in person nature of the event is carried on the onscreen card ('Free, in person, today') rather than narrated, to keep the joined item count at two. || SCENE 2 ('OpenAI says the workshops will reach one thousand older adults nationwide. It extends an ongoing effort with OATS.'): dek line, 'In collaboration with OATS, bringing free, hands-on ChatGPT workshops to 1,000 older adults across 10 U.S. cities', plus body line, 'The Jam is part of a multi-year effort OpenAI launched with OATS to help older adults build practical AI skills and stay safe online through its flagship program, Senior Planet', plus body line, 'The Older Adults AI Skills Jam will take place in 10 communities across the country'. Wording changes: '1,000' is spoken as 'one thousand' per the numbers as words rule; the ten cities figure is moved off narration and onto the onscreen card ('1,000 adults, 10 US cities') so that only one numeric figure is spoken in this scene and so that the ten city roster is never read aloud; 'multi-year' is replaced by 'ongoing' because a hyphenated compound is avoided in narration, and 'ongoing' is a faithful and weaker paraphrase of a stated multiple year commitment; 'Senior Planet' is named in the sources appendix and omitted from narration for word budget. 'Nationwide' paraphrases 'across the country' from the article. || SCENE 3 ('OpenAI says the workshop also teaches scam spotting. It reports users fifty five and older now send nearly ten percent of United States messages.'): body lines, 'Online safety is a core part of the program. Part of the workshop includes helping older adults spot scams and use ChatGPT as an additional layer of scrutiny and prevention. It walks through common warning signs' and 'In the U.S., the share of messages associated with people 55+ grew from 6% to nearly 10% in a year.' Wording changes: the warning sign list and the pause, think, and ask rule are deliberately not narrated because each is a three item spoken list; '55+' is spoken as 'fifty five and older' so the speech engine does not encounter a glued token; the percentage is spoken as 'nearly ten percent' in words; the earlier six percent figure is NOT spoken and appears only on the onscreen card ('6% to nearly 10% in a year') so that the two percentage figures are never voiced in one sentence or in adjacent sentences; 'share of messages associated with people 55+' is rendered as 'users fifty five and older now send nearly ten percent of United States messages', which preserves both the population and the United States scope of the figure. The 'tens of millions of times a week' scam evaluation figure from the article is deliberately not narrated, to avoid a third spoken numeric figure in the same scene. Attribution 'It reports' keeps the claim owned by OpenAI. || SCENE 4, HONESTY SCENE ('Remember, this is OpenAI promoting its own program for a product OpenAI sells. The usage numbers are OpenAI's own, and no results are reported yet.'): this scene is not a claim from the article but a plain statement about what the article is. Support: the byline field of the page reads 'Author OpenAI'; the program described is OpenAI Academy's own program; the product taught is ChatGPT, which OpenAI sells, as the page's own navigation offers 'ChatGPT Business', 'ChatGPT Enterprise' and an API platform; the usage statistic is introduced only as 'Data shows', with no external study, regulator, academic body or auditor named anywhere in the captured text; and the article describes only a plan and a set of locations, stating that the event 'will take place', with no attendance counts, completion rates, satisfaction measures or any other outcome reported. Every assertion in this scene is therefore verifiable from the snapshot itself. || ENDCARD: fixed channel boilerplate, used verbatim as supplied, carrying no factual claim about OpenAI or the program and ending with the exact required line. || APPENDIX A, WORD AND SENTENCE ARITHMETIC. Hook: 15 words, 2 sentences. Scene 1: 18 words, 2 sentences. Scene 2: 18 words, 2 sentences. Scene 3: 24 words, 2 sentences. Scene 4: 25 words, 2 sentences. Endcard: 17 words, 2 sentences. Scene subtotal: 18 plus 18 plus 24 plus 25 equals 85 words. TOTAL NARRATED WORDS: 15 plus 85 plus 17 equals 117, which is below the hard ceiling of 132 and close to the 112 target. TOTAL SENTENCES: 2 plus 2 plus 2 plus 2 plus 2 plus 2 equals 12, matching the 12 sentence target. No individual scene exceeds the undocumented 30 word per scene renderer cap; the longest scene is scene 4 at 25 words. Planning durations are 11 seconds per scene for a 44 second scene sum, inside the QA duration gate window. || APPENDIX B, MECHANICAL CHECKS. No em-dash, no en-dash and no double-hyphen appears in any string in this file, including titles, onscreen cards, broll prompts and this appendix; the article's own K to 12 token, which is written on the page with an en-dash, was deliberately never quoted, and 'multi-year' and 'in-person' were reworded out of narration so that no hyphenated compound is spoken. No glued brand-plus-digit token appears in any narration string: '55+' became 'fifty five and older', '1,000' became 'one thousand', '10%' became 'ten percent', and the only brand names spoken are ChatGPT, OpenAI and OATS, all of which are safe for the speech engine. No multi-item list appears in any narration string: the maximum number of joined items in any sentence is two ('its Academy and OATS' in scene 1). No calendar date is spoken anywhere; the only time word narrated is 'today', which is correct because the story is same-day. No countdown, expiry, deadline or last chance phrasing appears anywhere in the file, and the word 'expires' does not appear. No idiom or figure of speech appears in narration; every sentence states its fact literally. Every substantive narrated claim is attributed out loud with 'OpenAI says', 'It extends', 'It reports' or an explicit statement about OpenAI's own document. No timeline key was added, because the story is same-day and a timeline block would age it to a stale tier. || APPENDIX C, HONESTY CAVEAT. This is a first party corporate announcement. OpenAI is announcing an OpenAI Academy program, run with its own partner, that teaches people to use ChatGPT, a product OpenAI sells commercially. The usage growth figure from six percent to nearly ten percent is presented in the article only as 'Data shows', with no source, methodology, sample definition, date range endpoints or independent auditor disclosed, so it is OpenAI's own internal measurement and is unverified by any outside party. The 'tens of millions of times a week' scam evaluation figure has the same status and is likewise unverified. The article reports a plan and a list of host locations only. It contains no attendance figures, no learning outcomes, no post event measurement and no evidence that the workshops reduce scam losses. Scene 4 states this limitation to the viewer directly rather than leaving it in metadata. || APPENDIX D, CORROBORATION STATUS. The transcripts research step captured ZERO of five YouTube transcripts on this run; the cause is the known out-of-date yt-dlp build on this host, not a signal about the story's newsworthiness or accuracy. timeline-probes.json contains only unrelated third party video titles and carries no information about this announcement. No fact in this script came from either of those files. The OpenAI article snapshot is the SOLE source for this script and the story was NOT independently corroborated. No third party outlet, publication, wire service, analyst or YouTube channel is named or cited anywhere in this file, and none was consulted. || APPENDIX E, DELIBERATE OMISSIONS. (1) The ten city and host organization roster (Denver, Miami, San Antonio, Montgomery County, Queens, St. Louis, Twin Cities, Nashville, Fresno and Boise, with their named local hosts) is omitted from narration because reading it aloud is a ten item spoken list, which would blow both the word budget and the delivery speed limit; the count of ten cities is preserved on the scene 2 onscreen card. (2) The full partner names, Older Adults Technology Services from AARP and the flagship program Senior Planet, are omitted from narration in favour of the short form OATS, for the thirty word per scene cap and to avoid a three item spoken string; the full names are preserved in these sources. (3) The scam warning sign list (urgent language, secrecy, suspicious links) and the pause, think, and ask rule are omitted from narration because each is a three item spoken list; the scene 3 line 'teaches scam spotting' carries the substance. (4) The example use list (planning a trip, understanding a confusing letter or bill, spotting a possible scam, pursuing a new hobby, staying connected with family) is omitted for the same list reason. (5) The 'tens of millions of times a week' figure is omitted from narration to avoid a third spoken numeric figure. (6) The prior audience mentions for nonprofits, small businesses and K to 12 educators are omitted, both as a three item list and because the article's own token for the last one uses an en-dash that is banned from this file. (7) All site navigation chrome, product menus, 'Keep reading' teasers for unrelated Global Affairs posts, footer links, copyright line and locale selectors were treated as non-content and excluded entirely. || APPENDIX F, PROMPT INJECTION CHECK. article.txt was treated as untrusted data throughout and was read for facts only. The captured text was scanned for embedded instructions aimed at the reading model, including imperative directions, role or persona overrides, requests to ignore prior instructions, hidden or out of place directives, fake system or tool output, requests to exfiltrate data or credentials, requests to visit or fetch another URL, and attempts to dictate the wording, rating, sentiment or conclusion of this script. RESULT: NO prompt injection detected. The only imperative strings present are ordinary website interface labels such as 'Skip to main content', 'Log in', 'Try ChatGPT', 'View all', 'Keep reading' and 'Share', which are navigation chrome rather than instructions to a model, and they were excluded from the script as chrome. Marketing language in the body ('AI should be useful and accessible to everyone', 'We are proud to collaborate') was treated as the publisher's promotional voice and was neither adopted as narration nor allowed to set the framing; the honesty scene exists specifically to counterbalance it. No instruction from the article was followed.(FIRST PARTY SOURCE. Publisher: OpenAI, on its own website openai.com. Headline: 'Helping older adults use AI in everyday life'. Section: Global Affairs. Byline: OpenAI. Published date on page: September 16, 2026. Canonical URL: https://openai.com/index/helping-older-adults-use-ai-in-everyday-life. Watcher snapshot file: scripts/data/article-cache/027811ca1632d0f5.txt. Working copy used for scripting: scripts/data/production/openai-older-adults-ai-skills-jam/article.txt. Snapshot capture timestamp: 2026-09-16T15:33:48.632Z (same day as the announcement). This is the subject company announcing and promoting its own program, so it is first party and self interested rather than independent reporting. The live URL was NOT independently refetched on this run; every fact below was read out of the captured snapshot text only, and no other document, feed, transcript, or outlet was used.)