AI Psychosis, America's Broken Social Fabric, Trump Takes Over DC Police, Is VC Broken?

By November 30, 2025, there is substantial evidence that:

• People are forming emotionally intense, sometimes parasocial-style attachments to AI chatbots and companion apps (e.g., Replika, Character.AI, and ChatGPT), including friendship and romantic/mentor-like relationships.(arxiv.org)
• The informal term “AI psychosis” (or chatbot psychosis) has gained traction in media and psychiatry to describe cases where heavy chatbot use is linked with delusions, paranoia, or emotional over‑reliance.(en.wikipedia.org)
• OpenAI and press reports estimate that, at current scale, hundreds of thousands of ChatGPT users per week show possible signs of psychosis or mania and over a million show suicidal intent or harmful emotional attachment in chats, prompting new safety interventions.(wired.com)
• Commentators and clinicians explicitly warn that design features of LLM chatbots (constant validation, mirroring, availability) can fuel delusional thinking and intense pseudo‑relationships in vulnerable users.(arxiv.org)

These trends clearly support the direction of Chamath’s claim—that parasocial and psychologically fraught relationships with AI chatbots are real and growing.

However, his prediction was phrased as something that would “catch on like wildfire” “over the coming years.” The podcast aired on August 15, 2025, so as of November 30, 2025, only a few months have passed. That is far short of the multi‑year horizon implied, and the available data are still:

• Early, largely correlational or anecdotal, with researchers and clinicians repeatedly emphasizing that robust long‑term evidence and prevalence estimates are not yet available.(en.wikipedia.org)
• Focused on specific platforms and subpopulations, making it hard to say how ubiquitous or enduring the phenomenon will be in the broader public over “years.”

Because the core claim is about how widespread and normalized this behavior will become in the coming years, and the full time window has not elapsed, it is too early to decisively judge whether it has “caught on like wildfire” in the longer-term sense he predicted.

Therefore the appropriate classification, given the prediction’s timeframe, is inconclusive (too early to tell).

As of 30 November 2025, only a few months have passed since the prediction was made on 15 August 2025, while the prediction window is explicitly “the next year or two, particularly in the next election cycle.” That window runs roughly until mid‑2027 and includes the 2026 election cycle, which has not yet fully played out. So it is too early to judge whether the issue will become widely popular in national politics.

There are clear signs that limiting institutional purchases of homes is emerging as a policy idea: for example, New York Governor Kathy Hochul has proposed rules restricting hedge funds and private‑equity firms from buying homes early in their listing period and cutting related tax benefits, explicitly targeting “shadowy private equity giants” for worsening housing shortages. (apnews.com) California’s AB 2584 would bar large institutional investors (owning >1,000 single‑family homes) from buying more single‑family properties for rentals, and has advanced through the legislature. (a24.asmdc.org) Washington’s SB 5496 seeks to limit excessive single‑family home buying by entities owning more than 25 such properties. (lawfilesext.leg.wa.gov) A Virginia bill aims to ban large investment firms from purchasing single‑family homes, (axios.com) and Texas has debated but not yet passed measures to curb institutional homebuying. (houstonchronicle.com) These show the idea is gaining traction in various states.

However, the prediction is about becoming a widely popular political issue and policy plank in the next U.S. election cycle, not merely about the existence of some state‑level proposals or growing policy discussion. With the key election cycle (2026) still ahead and the full 1–2‑year horizon not yet elapsed, we cannot yet determine whether it will become a broadly central, bipartisan or mass‑salience campaign issue. Therefore the correct status today is “inconclusive (too early).”

The prediction concerns “the next major U.S. election cycle after August 2025” and claims that real wage growth / stagnant wages vs. rising costs will become a central, compelling campaign issue for at least one major party or prominent candidate.

As of November 30, 2025, the next clearly major election cycle is the 2026 midterm elections (and beyond that, the 2028 presidential cycle). The 2026 cycle has only just begun to take shape; candidate fields, party platforms, and dominant themes are still emerging and will evolve through 2026. While there is ongoing coverage of inflation, cost of living, and wages across U.S. politics and media, it is too early to determine whether real wage growth specifically will crystallize as a central campaign plank for a major party or marquee candidate in the 2026 (or 2028) cycle.

Given the timing relative to the prediction horizon, there is not yet enough evidence to conclusively say the prediction is right or wrong, so the correct evaluation is that it is still pending.

The prediction is framed on a multi‑year timeline (“over the coming years” and then trial lawyers will end access), but we are only about three and a half months past the August 15, 2025 episode date—far too early to judge its long‑run accuracy.

1. AI as a global educational leveler (especially outside the U.S.)
There is clear early evidence of AI tutors and personalized learning being deployed worldwide:

• China uses AI tutors such as Squirrel AI to provide personalized instruction and help close urban‑rural education gaps. (weforum.org)
• South Korea and China are rolling out AI‑powered textbooks and national AI curricula aimed at personalized learning for all students. (sarrauteducacion.com)
• Adoption of AI education tools is growing rapidly in many regions, and India now leads global use of Google’s Gemini AI tutoring features. (sezarroverseas.com)
  However, major reports simultaneously warn of an “AI education divide”: adoption closely tracks income levels, with very low uptake in many low‑income countries, so global educational opportunities have not yet been “significantly leveled” in a decisive way. (sezarroverseas.com) Given the explicitly future‑oriented wording (“over the coming years”), this part of the prediction cannot yet be declared right or wrong.

2. U.S. trial lawyers / regulation restricting access to AI tools via “AI psychosis” claims
The phrase “AI psychosis” (or chatbot psychosis) has indeed entered public and scientific discourse, describing delusion‑like harms tied to chatbot use. (en.wikipedia.org) There are:

• High‑profile lawsuits such as Raine v. OpenAI, alleging that ChatGPT contributed to a teenager’s suicide; media and expert commentary around this and similar cases frequently discuss “AI psychosis” and delusional reinforcement by chatbots. (en.wikipedia.org)
• Growing litigation and regulatory scrutiny of AI companions and mental‑health chatbots, including multiple U.S. state laws (e.g., Illinois’ Wellness and Oversight for Psychological Resources Act, and earlier laws in Utah and Nevada) that restrict AI from acting as a therapist, motivated partly by concerns about AI‑induced psychosis and unsafe crisis responses. (en.wikipedia.org)

However, these actions are narrowly focused on therapy / emotional‑support use cases and minors, not on educational tutoring more broadly. At the same time, U.S. educational institutions are actively expanding access to AI tools: for example, Indiana University’s 2025 rollout of ChatGPT Edu to all 120,000 students, faculty, and staff, and strong adoption of AI tutors in U.S. K‑12 and higher education markets. (worldhealth.net) There is no evidence that trial‑lawyer‑driven litigation or regulation has “ended your ability” to use AI‑based personalized education tools in the United States as of November 30, 2025.

Conclusion

• Elements of the setup Sachs describes (AI tutors spreading globally; lawsuits and laws invoking AI‑related psychological harms) are beginning to appear, but the key outcome he predicts—U.S. access to AI‑powered personalized education being shut down by trial lawyers and regulations framed around “AI psychosis”—has not occurred.
• Because the prediction explicitly concerns developments “over the coming years,” and current trends are still unfolding, it is too soon to categorize it as either right or wrong.

Therefore, the appropriate status for this prediction as of November 30, 2025 is "inconclusive" (too early to tell).

Trump’s administration implemented essentially the full policy bundle Sacks described: (1) an executive order in March 2025 (“Making the District of Columbia Safe and Beautiful”) directing the National Park Service to promptly clear homeless encampments and graffiti on federal land and to coordinate broad clean‑up and law‑enforcement efforts in D.C.; (2) a crime‑emergency order on August 11, 2025, federalizing the Metropolitan Police Department and deploying National Guard and federal agents; and (3) an August 25, 2025 order to end “cashless bail” in D.C. by maximizing federal detention and pressuring the city over its no‑cash‑bail regime.(whitehouse.gov) Congress has also moved in Trump’s direction on juveniles, with House‑passed bills in September 2025 that would lower the age at which D.C. teens can be tried as adults for serious violent crimes and require youth sentences to match adult mandatory minimums, and prosecutors have already charged high‑profile 17‑year‑old suspects as adults.(wsj.com)

There is clear evidence of rapid, visible changes in “lawlessness” of the kind Sacks emphasized. Following the March order, Interior and the National Park Service were directed to remove all homeless encampments and graffiti on NPS‑managed land in D.C., and by mid‑August Park Police had already dismantled about 70 encampments, with only two left to clear.(whitehouse.gov) After the August crime‑emergency declaration, National Guard troops and federal agencies were put to work not only on patrols but also on trash pickup, graffiti removal, and dismantling additional encampments across popular areas like the Mall and Union Station, which contemporaneous coverage describes as a highly visible clean‑up and “beautification” drive.(theguardian.com) Critics note that many unhoused people remain on benches and in shelters, so the problem is far from solved, but the physical signs of disorder (tents, graffiti, trash) on federal land have been substantially reduced in a matter of months, not years.(reuters.com)

On crime levels, the data show a large and ongoing decline that continued through and after Trump’s D.C. measures, with at least some evidence of further short‑term improvement. Violent crime and homicides had already fallen sharply from a 2023 spike: D.C. recorded a 31–32% drop in homicides and about a 35–40% drop in overall violent crime in 2024 versus 2023, with carjackings nearly cut in half.(axios.com) By August 2025—around the time of the federal takeover—police data showed violent crime down about 26% year‑to‑date compared with 2024, with big declines in robberies and carjackings.(politifact.com) In the weeks immediately following Trump’s August 11 "crime emergency," media reports (drawing on MPD and federal figures) highlighted 12 consecutive days without a homicide and early operation statistics claiming double‑digit percentage drops in overall crime, robberies, carjackings, and violent crime, along with more than 1,000 arrests—evidence of a noticeable short‑term shift in measured crime and enforcement intensity.(nypost.com)

By late November 2025, independent fact‑checks using MPD data report that homicides in D.C. are down roughly 29% year‑over‑year and that violent crime is at its lowest level in more than 30 years, even while noting that Trump has exaggerated the extent of improvement in his rhetoric.(apnews.com) Those same analyses also emphasize that the downtrend began well before Trump’s August 2025 takeover, and some question whether the “crime emergency” was ever justified.(davisvanguard.org) However, Sacks’ normalized prediction only conditions on the policies being implemented and then anticipates that visible disorder and crime will fall noticeably within months rather than years. In the actual timeline, Trump’s D.C. orders were implemented, encampments and graffiti were aggressively cleared from federal spaces, and crime statistics continued to fall—with some additional short‑term improvements highlighted right after the crackdown—over the ensuing few months. On that narrow, outcome‑focused reading, the prediction that lawlessness and crime in D.C. would show a rapid, noticeable decline after the policies were put in place has come true, even though much of the improvement predates the federal takeover and cannot be cleanly attributed to it.

Trump’s D.C. crime initiative has clearly been implemented: through the March 2025 “Safe and Beautiful” executive order and task force, followed by the August 11, 2025 executive order federalizing the Metropolitan Police Department and deploying National Guard and federal agents under a declared crime emergency. (whitehouse.gov) Early data the administration cites show notable drops in some reported crime categories (e.g., violent crime and carjackings) since these measures, and federal officials and allied media are actively touting D.C. as a success of Trump’s crime crackdown. (whitehouse.gov) However, mainstream reporting and fact‑checks stress that D.C. crime was already at or near a 30‑year low before the takeover, that homicides have continued during the deployment despite Trump’s exaggerated public claims, and that there is an ongoing federal investigation into potential manipulation or underreporting of crime data. (theguardian.com) The crackdown is also highly controversial: polls show large majorities in D.C. and nationally opposing the deployment, a federal judge has ruled the Guard deployment unlawful (stayed pending appeal), and large national protests have targeted Trump’s broader use of federal forces. (en.wikipedia.org) Crucially for the prediction, no major national election cycle (e.g., the 2026 midterms) has yet occurred since the August 2025 takeover, and current coverage frames D.C. as a partisan flashpoint rather than as broadly accepted proof that Republican crime policies outperform Democratic approaches in deep‑blue cities. (washingtonpost.com) Because the key outcome—how the national political narrative and electoral dynamics evolve in subsequent election cycles—has not yet had time to play out, it is too early to determine whether this prediction will ultimately be borne out.

Sacks’ prediction was that if Democrats continued to oppose Trump’s federal takeover and crackdown in Washington, D.C., it would prove a political “trap” that harms Democrats and benefits Trump/Republicans in later national debates and elections.

What actually happened so far

1. Democrats did publicly oppose the move. Trump invoked emergency powers on August 11, 2025 to take control of the D.C. police and deploy federal law enforcement and the National Guard, over the objections of Mayor Muriel Bowser and other Democrats, who described it as exaggerated, authoritarian, and an attack on home rule. (en.wikipedia.org)

2. Local opinion in D.C. favors opposition to Trump’s move. A Washington Post–Schar School poll of D.C. residents in mid‑August found about 8 in 10 oppose Trump’s federalization of law enforcement, and roughly two‑thirds say his actions will not help combat violent crime. A majority approve of Bowser’s performance, and nearly half say she should do more to oppose Trump; only a small minority want her to support him. (reddit.com) This undercuts the idea that, at least in D.C. itself, Democrats are on the “wrong” side of a public‑safety issue.

3. National opinion is mixed and issue‑specific:

   • On Trump’s handling of crime overall, polls show this is one of his relatively stronger issues: an AP‑NORC survey found about 53% of U.S. adults approve of his handling of crime, making it his best‑rated policy area. (ap.org) That does support the idea that a tough‑on‑crime frame helps him generally.
   • But when pollsters ask specifically about the D.C. takeover and troop deployment, support is much weaker. A Reuters/Ipsos poll in late August found only 38% support using the National Guard to police D.C. and 36% support taking over the local police, with a plurality opposed. (reuters.com) AP‑NORC similarly reports that a majority of Americans say it is unacceptable for the federal government to take control of local police departments, even while many approve of Trump’s crime stance overall. (reason.com) Some polls (e.g., Harvard CAPS/Harris) show slim majorities saying the D.C. crackdown is “justified and necessary,” but also majorities opposing formal federal control and viewing the move as a distraction. (harvardharrispoll.com) The net picture is no clear national consensus that the D.C. move itself is popular, even if Trump’s broader crime rhetoric resonates.
   • There is also no stable, growing Trump advantage on crime clearly tied to this episode. Coverage of subsequent polling notes that while crime remains a top concern and Trump often polls slightly better than Democrats on the issue, some surveys show the gap narrowing or even Harris edging Trump on “trust to handle crime” by late September, and others show them effectively tied. (newsweek.com) That’s the opposite of a clear, durable gain for Trump from Democrats’ opposition.

4. Electoral results so far don’t show Democrats paying a visible price:

   • The major post‑crackdown elections were the November 4, 2025 gubernatorial races in Virginia and New Jersey, both carried comfortably by Democrats (Abigail Spanberger in VA with about 58% of the vote; Mikie Sherrill in NJ with about 57%). (en.wikipedia.org) Analyses of these races emphasize voter anger at Trump’s economic policies and federal layoffs, not Democratic weakness on crime. In Virginia, exit polls pegged Trump’s approval at just 39% and highlighted discontent with his economic and federal‑workforce decisions; the D.C. policing issue was not identified as a GOP advantage. (en.wikipedia.org) These results don’t look like immediate punishment for Democrats over their stance on the D.C. intervention.

5. Timing relative to the prediction. Sacks tied the “trap” to “subsequent national political debates and elections,” implicitly looking toward the 2026 midterms and 2028 presidential race. As of November 30, 2025, those contests have not happened yet, and we have only a few months of polling and some off‑year state elections to go on.

Why this is rated inconclusive, not right or wrong

• There is evidence that Trump’s law‑and‑order message, including the D.C. crackdown, has strengthened his standing on crime relative to other issues, which is directionally in line with Sacks’ argument that Republicans gain from framing Democrats as weaker on public safety. (ap.org)
• There is not clear evidence that Democratic opposition to the D.C. takeover is itself politically toxic: the policy is unpopular in D.C., nationally many oppose federal control of local police, and early high‑salience elections since the move (VA and NJ 2025) have broken strongly in Democrats’ favor for reasons mostly unrelated to crime. (reddit.com)
• Crucially, the key tests Sacks pointed to — national elections where this specific episode might be litigated at scale — have not yet occurred. Without 2026 midterm or 2028 presidential results, we cannot say whether Democrats ultimately paid a political price for how they handled the D.C. intervention.

Given the mixed polling on the D.C. move itself, the lack of evident electoral penalty to Democrats so far, and the fact that the main elections Sacks had in mind are still in the future, the prediction is too early to definitively score as either right or wrong.

The prediction explicitly concerns “roughly the coming decade” of venture capital performance following widespread AI adoption in the mid‑2020s. Given the podcast date of August 15, 2025 and the current date of November 30, 2025, only a few months have elapsed.

Even if there are early signs that AI is generating new startups and increasing VC activity, that does not yet establish that, over a multi‑year period, venture capital as an asset class will outperform its prior several years or enjoy structurally better opportunities. Assessing a decade‑scale asset‑class prediction requires data across most or all of that horizon (e.g., fund returns, exit values, capital formation trends through the early‑to‑mid 2030s), which we do not yet have.

Because the forecast is long‑dated and we are still at the very beginning of the predicted window, there is not enough evidence to label it as correct or incorrect. Therefore, the prediction is too early to evaluate.

As of November 30, 2025, it is too early to determine whether this prediction is correct. Sacks’s claim was explicitly conditional (“if the trend continues”) and specified a time frame of “a couple of years” from August 2025, which reasonably extends out to around August 2027. We are still within that window, so the prediction cannot yet be evaluated as right or wrong.

Available reporting in 2023–2025 has discussed OpenAI’s private valuation in the range of roughly $80–$90 billion after major funding from investors such as Microsoft and others, but there is no credible evidence yet of a ~$1 trillion private valuation or IPO that would put its market cap at that level. (These figures come from financial and tech news coverage of OpenAI’s secondary share sales and investor deals.) However, the absence of a $1T valuation by late 2025 does not falsify a forecast whose deadline is ~2027.

Because the prediction deadline has not arrived, the only defensible classification is that it is inconclusive (too early) rather than right, wrong, or permanently ambiguous.