Ledger // 27 claims on record
Predictions,
graded
Anyone can sound clever about the future. The test is whether you were right, at the confidence you claimed, by the date you named. Each claim links back to the edition that made it. Resolved claims are never deleted.
- Right
- 0
- Wrong
- 0
- Partial
- 0
- Open
- 27
Counted on the day this page was last built (2026-11-11). Nothing here updates itself.
The horizon // when each claim comes due
Every claim, placed on a line at the date by which it must be settled. The rule is today. As claims resolve, the marks change colour and the rule moves past them.
Calibration // 0 resolved · 27 open
Spread // mean 56%
How confident the claims are, before any of them are graded. A writer who hedges everything at fifty is saying nothing; one who calls everything at ninety is about to be embarrassed.
- 25%1
- 30%2
- 40%1
- 45%3
- 50%1
- 55%3
- 60%6
- 65%4
- 70%4
- 75%2
Due by year
- 20276
- 202810
- 20296
- 20304
- 20311
Every claim
| Ed. | Claim | Confidence | Resolve by | Status |
|---|---|---|---|---|
| 009 | By the end of 2027, the United States will still have no federal statute creating a right in one's own voice or likeness against A.I. replication; the NO FAKES Act or a successor will not have been signed into law. ¶ | 65% | 2027-12-31 | open |
| 009 | By the end of 2028, neither the United Kingdom nor Canada will have enacted a statutory right of publicity for living persons covering A.I. voice replication. ¶ | 60% | 2028-12-31 | open |
| 009 | The Presley and Eight Mile suits against Suno will end in settlement or licence rather than a judgment on the merits of the training claims, by the end of 2028. ¶ | 70% | 2028-12-31 | open |
| 008 | By the end of 2027, at least one G7 country will have enacted a statute or binding regulation setting standards for A.I. chatbots' responses to election questions (accuracy, sourcing, disclosure or refusal), beyond voluntary codes. ¶ | 45% | 2027-12-31 | open |
| 008 | Before the next Canadian or British general election, whichever comes first, a national regulator or electoral authority in that country will publish a pre-election audit of chatbot voting advice. ¶ | 60% | 2029-12-31 | open |
| 008 | By the 2028 U.S. presidential election, at least one of the three most-used consumer chatbots will name a preferred candidate in a general-election race when asked plainly, without a jailbreak, as default behaviour. ¶ | 30% | 2028-11-07 | open |
| 007 | By the end of 2027, the U.K. will have no primary legislation specifically governing police use of live facial recognition; deployments will continue under the existing guidance-and-case-law framework. ¶ | 65% | 2027-12-31 | open |
| 007 | By the end of 2028, at least one Canadian province will restrict police use of live facial recognition or automated licence-plate readers by statute or binding regulation, beyond a moratorium or guideline. ¶ | 40% | 2028-12-31 | open |
| 007 | By the end of 2029, at least one G7 government will have formally disclosed that a surveillance or targeting system it procured was first deployed by another state in an armed conflict, as a condition of procurement rather than after a journalist found out. ¶ | 25% | 2029-12-31 | open |
| 006 | By the end of 2027, at least one of Tinder, Hinge or Bumble will ship a feature that marks or verifies messages as human-written, or discloses A.I. assistance, in at least one market. ¶ | 45% | 2027-12-31 | open |
| 006 | When the Institute for Family Studies/YouGov question is next repeated, the share of Americans under 40 who say A.I. could replace real-life romance will exceed 35 per cent (up from 25 in 2024). ¶ | 55% | 2028-12-31 | open |
| 006 | By the end of 2030, at least one G7 country will regulate A.I. companion apps for adults specifically — disclosure, dependency safeguards or design rules — rather than only for minors. ¶ | 60% | 2030-12-31 | open |
| 005 | By the end of 2028, no U.S. federal statute specifically regulating A.I. companions or chatbots for minors will have been enacted; the field will still be governed by state suits and consumer-protection law. ¶ | 65% | 2028-12-31 | open |
| 005 | By the end of 2027, at least one of this week's state suits against OpenAI will produce a settlement or consent decree that imposes age-verification or parental-control requirements on ChatGPT. ¶ | 55% | 2027-12-31 | open |
| 005 | By the end of 2029, a peer-reviewed study of children under 13 will report a measurable reduction in task persistence associated with regular A.I. assistant use, replicating the 2026 Berkeley adult finding. ¶ | 70% | 2029-12-31 | open |
| 004 | By 2030, labour's share of income in the United States (BLS nonfarm business measure) will be at least two percentage points below its 2024 level. ¶ | 60% | 2030-12-31 | open |
| 004 | By the end of 2031, at least one G7 country will have enacted a permanent, universal cash transfer or citizen dividend at national level, funded at least in part from taxation of automated production or from a public stake in A.I. or compute infrastructure. ¶ | 30% | 2031-12-31 | open |
| 004 | By the end of 2029, no publicly owned, co-operative or non-profit-governed model will rank in the top five on the major independent capability leaderboards. ¶ | 75% | 2029-12-31 | open |
| 003 | The reopened Tompkins County investigation will produce at least one criminal indictment by the end of 2027. ¶ | 50% | 2027-12-31 | open |
| 003 | By the end of 2028, the FTC will have publicly announced fewer than ten enforcement actions under the Take It Down Act. ¶ | 65% | 2028-12-31 | open |
| 003 | By the end of 2028, the Ofcom investigation into X over Grok's sexualized images will have produced no fine larger than £10 million. ¶ | 60% | 2028-12-31 | open |
| 002 | By the end of 2028, fewer than one in twenty U.S. private-sector workers will be covered by a collective agreement containing enforceable limits on an employer's use of generative A.I. ¶ | 75% | 2028-12-31 | open |
| 002 | By the end of 2029, no US federal law will impose liability on A.I. developers for harms caused by model outputs in ordinary commercial use. ¶ | 70% | 2029-12-31 | open |
| 002 | US union membership as a share of employed workers (BLS annual figure) will be lower for 2029 than for 2024. ¶ | 60% | 2030-01-31 | open |
| 001 | By 31 December 2028, job postings with the title 'prompt engineer' on the major English-language job boards will number fewer than one third of their 2024–25 peak. ¶ | 70% | 2028-12-31 | open |
| 001 | By the end of 2029, at least one of the five largest global consultancies will have dissolved its standalone A.I. practice back into a general technology practice. ¶ | 55% | 2029-12-31 | open |
| 001 | By the end of 2030, at least one G7 country will require a named, accountable human sign-off for A.I.-produced work in at least one licensed profession (law, accounting, medicine, engineering). ¶ | 45% | 2030-12-31 | open |
Variable // The capacity to resist · 9
| Ed. | Claim | Confidence | Resolve by | Status |
|---|---|---|---|---|
| 007 | By the end of 2027, the U.K. will have no primary legislation specifically governing police use of live facial recognition; deployments will continue under the existing guidance-and-case-law framework. ¶ | 65% | 2027-12-31 | open |
| 007 | By the end of 2028, at least one Canadian province will restrict police use of live facial recognition or automated licence-plate readers by statute or binding regulation, beyond a moratorium or guideline. ¶ | 40% | 2028-12-31 | open |
| 007 | By the end of 2029, at least one G7 government will have formally disclosed that a surveillance or targeting system it procured was first deployed by another state in an armed conflict, as a condition of procurement rather than after a journalist found out. ¶ | 25% | 2029-12-31 | open |
| 004 | By 2030, labour's share of income in the United States (BLS nonfarm business measure) will be at least two percentage points below its 2024 level. ¶ | 60% | 2030-12-31 | open |
| 004 | By the end of 2031, at least one G7 country will have enacted a permanent, universal cash transfer or citizen dividend at national level, funded at least in part from taxation of automated production or from a public stake in A.I. or compute infrastructure. ¶ | 30% | 2031-12-31 | open |
| 004 | By the end of 2029, no publicly owned, co-operative or non-profit-governed model will rank in the top five on the major independent capability leaderboards. ¶ | 75% | 2029-12-31 | open |
| 002 | By the end of 2028, fewer than one in twenty U.S. private-sector workers will be covered by a collective agreement containing enforceable limits on an employer's use of generative A.I. ¶ | 75% | 2028-12-31 | open |
| 002 | By the end of 2029, no US federal law will impose liability on A.I. developers for harms caused by model outputs in ordinary commercial use. ¶ | 70% | 2029-12-31 | open |
| 002 | US union membership as a share of employed workers (BLS annual figure) will be lower for 2029 than for 2024. ¶ | 60% | 2030-01-31 | open |
Variable // Labour scarcity · 3
| Ed. | Claim | Confidence | Resolve by | Status |
|---|---|---|---|---|
| 004 | By 2030, labour's share of income in the United States (BLS nonfarm business measure) will be at least two percentage points below its 2024 level. ¶ | 60% | 2030-12-31 | open |
| 004 | By the end of 2031, at least one G7 country will have enacted a permanent, universal cash transfer or citizen dividend at national level, funded at least in part from taxation of automated production or from a public stake in A.I. or compute infrastructure. ¶ | 30% | 2031-12-31 | open |
| 004 | By the end of 2029, no publicly owned, co-operative or non-profit-governed model will rank in the top five on the major independent capability leaderboards. ¶ | 75% | 2029-12-31 | open |
Variable // The interface gap · 3
| Ed. | Claim | Confidence | Resolve by | Status |
|---|---|---|---|---|
| 001 | By 31 December 2028, job postings with the title 'prompt engineer' on the major English-language job boards will number fewer than one third of their 2024–25 peak. ¶ | 70% | 2028-12-31 | open |
| 001 | By the end of 2029, at least one of the five largest global consultancies will have dissolved its standalone A.I. practice back into a general technology practice. ¶ | 55% | 2029-12-31 | open |
| 001 | By the end of 2030, at least one G7 country will require a named, accountable human sign-off for A.I.-produced work in at least one licensed profession (law, accounting, medicine, engineering). ¶ | 45% | 2030-12-31 | open |
Variable // Accountability · 21
| Ed. | Claim | Confidence | Resolve by | Status |
|---|---|---|---|---|
| 009 | By the end of 2027, the United States will still have no federal statute creating a right in one's own voice or likeness against A.I. replication; the NO FAKES Act or a successor will not have been signed into law. ¶ | 65% | 2027-12-31 | open |
| 009 | By the end of 2028, neither the United Kingdom nor Canada will have enacted a statutory right of publicity for living persons covering A.I. voice replication. ¶ | 60% | 2028-12-31 | open |
| 009 | The Presley and Eight Mile suits against Suno will end in settlement or licence rather than a judgment on the merits of the training claims, by the end of 2028. ¶ | 70% | 2028-12-31 | open |
| 008 | By the end of 2027, at least one G7 country will have enacted a statute or binding regulation setting standards for A.I. chatbots' responses to election questions (accuracy, sourcing, disclosure or refusal), beyond voluntary codes. ¶ | 45% | 2027-12-31 | open |
| 008 | Before the next Canadian or British general election, whichever comes first, a national regulator or electoral authority in that country will publish a pre-election audit of chatbot voting advice. ¶ | 60% | 2029-12-31 | open |
| 008 | By the 2028 U.S. presidential election, at least one of the three most-used consumer chatbots will name a preferred candidate in a general-election race when asked plainly, without a jailbreak, as default behaviour. ¶ | 30% | 2028-11-07 | open |
| 007 | By the end of 2027, the U.K. will have no primary legislation specifically governing police use of live facial recognition; deployments will continue under the existing guidance-and-case-law framework. ¶ | 65% | 2027-12-31 | open |
| 007 | By the end of 2028, at least one Canadian province will restrict police use of live facial recognition or automated licence-plate readers by statute or binding regulation, beyond a moratorium or guideline. ¶ | 40% | 2028-12-31 | open |
| 007 | By the end of 2029, at least one G7 government will have formally disclosed that a surveillance or targeting system it procured was first deployed by another state in an armed conflict, as a condition of procurement rather than after a journalist found out. ¶ | 25% | 2029-12-31 | open |
| 005 | By the end of 2028, no U.S. federal statute specifically regulating A.I. companions or chatbots for minors will have been enacted; the field will still be governed by state suits and consumer-protection law. ¶ | 65% | 2028-12-31 | open |
| 005 | By the end of 2027, at least one of this week's state suits against OpenAI will produce a settlement or consent decree that imposes age-verification or parental-control requirements on ChatGPT. ¶ | 55% | 2027-12-31 | open |
| 005 | By the end of 2029, a peer-reviewed study of children under 13 will report a measurable reduction in task persistence associated with regular A.I. assistant use, replicating the 2026 Berkeley adult finding. ¶ | 70% | 2029-12-31 | open |
| 003 | The reopened Tompkins County investigation will produce at least one criminal indictment by the end of 2027. ¶ | 50% | 2027-12-31 | open |
| 003 | By the end of 2028, the FTC will have publicly announced fewer than ten enforcement actions under the Take It Down Act. ¶ | 65% | 2028-12-31 | open |
| 003 | By the end of 2028, the Ofcom investigation into X over Grok's sexualized images will have produced no fine larger than £10 million. ¶ | 60% | 2028-12-31 | open |
| 002 | By the end of 2028, fewer than one in twenty U.S. private-sector workers will be covered by a collective agreement containing enforceable limits on an employer's use of generative A.I. ¶ | 75% | 2028-12-31 | open |
| 002 | By the end of 2029, no US federal law will impose liability on A.I. developers for harms caused by model outputs in ordinary commercial use. ¶ | 70% | 2029-12-31 | open |
| 002 | US union membership as a share of employed workers (BLS annual figure) will be lower for 2029 than for 2024. ¶ | 60% | 2030-01-31 | open |
| 001 | By 31 December 2028, job postings with the title 'prompt engineer' on the major English-language job boards will number fewer than one third of their 2024–25 peak. ¶ | 70% | 2028-12-31 | open |
| 001 | By the end of 2029, at least one of the five largest global consultancies will have dissolved its standalone A.I. practice back into a general technology practice. ¶ | 55% | 2029-12-31 | open |
| 001 | By the end of 2030, at least one G7 country will require a named, accountable human sign-off for A.I.-produced work in at least one licensed profession (law, accounting, medicine, engineering). ¶ | 45% | 2030-12-31 | open |
Variable // The cost of production · 15
| Ed. | Claim | Confidence | Resolve by | Status |
|---|---|---|---|---|
| 009 | By the end of 2027, the United States will still have no federal statute creating a right in one's own voice or likeness against A.I. replication; the NO FAKES Act or a successor will not have been signed into law. ¶ | 65% | 2027-12-31 | open |
| 009 | By the end of 2028, neither the United Kingdom nor Canada will have enacted a statutory right of publicity for living persons covering A.I. voice replication. ¶ | 60% | 2028-12-31 | open |
| 009 | The Presley and Eight Mile suits against Suno will end in settlement or licence rather than a judgment on the merits of the training claims, by the end of 2028. ¶ | 70% | 2028-12-31 | open |
| 008 | By the end of 2027, at least one G7 country will have enacted a statute or binding regulation setting standards for A.I. chatbots' responses to election questions (accuracy, sourcing, disclosure or refusal), beyond voluntary codes. ¶ | 45% | 2027-12-31 | open |
| 008 | Before the next Canadian or British general election, whichever comes first, a national regulator or electoral authority in that country will publish a pre-election audit of chatbot voting advice. ¶ | 60% | 2029-12-31 | open |
| 008 | By the 2028 U.S. presidential election, at least one of the three most-used consumer chatbots will name a preferred candidate in a general-election race when asked plainly, without a jailbreak, as default behaviour. ¶ | 30% | 2028-11-07 | open |
| 006 | By the end of 2027, at least one of Tinder, Hinge or Bumble will ship a feature that marks or verifies messages as human-written, or discloses A.I. assistance, in at least one market. ¶ | 45% | 2027-12-31 | open |
| 006 | When the Institute for Family Studies/YouGov question is next repeated, the share of Americans under 40 who say A.I. could replace real-life romance will exceed 35 per cent (up from 25 in 2024). ¶ | 55% | 2028-12-31 | open |
| 006 | By the end of 2030, at least one G7 country will regulate A.I. companion apps for adults specifically — disclosure, dependency safeguards or design rules — rather than only for minors. ¶ | 60% | 2030-12-31 | open |
| 004 | By 2030, labour's share of income in the United States (BLS nonfarm business measure) will be at least two percentage points below its 2024 level. ¶ | 60% | 2030-12-31 | open |
| 004 | By the end of 2031, at least one G7 country will have enacted a permanent, universal cash transfer or citizen dividend at national level, funded at least in part from taxation of automated production or from a public stake in A.I. or compute infrastructure. ¶ | 30% | 2031-12-31 | open |
| 004 | By the end of 2029, no publicly owned, co-operative or non-profit-governed model will rank in the top five on the major independent capability leaderboards. ¶ | 75% | 2029-12-31 | open |
| 001 | By 31 December 2028, job postings with the title 'prompt engineer' on the major English-language job boards will number fewer than one third of their 2024–25 peak. ¶ | 70% | 2028-12-31 | open |
| 001 | By the end of 2029, at least one of the five largest global consultancies will have dissolved its standalone A.I. practice back into a general technology practice. ¶ | 55% | 2029-12-31 | open |
| 001 | By the end of 2030, at least one G7 country will require a named, accountable human sign-off for A.I.-produced work in at least one licensed profession (law, accounting, medicine, engineering). ¶ | 45% | 2030-12-31 | open |
Variable // Institutional trust · 9
| Ed. | Claim | Confidence | Resolve by | Status |
|---|---|---|---|---|
| 009 | By the end of 2027, the United States will still have no federal statute creating a right in one's own voice or likeness against A.I. replication; the NO FAKES Act or a successor will not have been signed into law. ¶ | 65% | 2027-12-31 | open |
| 009 | By the end of 2028, neither the United Kingdom nor Canada will have enacted a statutory right of publicity for living persons covering A.I. voice replication. ¶ | 60% | 2028-12-31 | open |
| 009 | The Presley and Eight Mile suits against Suno will end in settlement or licence rather than a judgment on the merits of the training claims, by the end of 2028. ¶ | 70% | 2028-12-31 | open |
| 008 | By the end of 2027, at least one G7 country will have enacted a statute or binding regulation setting standards for A.I. chatbots' responses to election questions (accuracy, sourcing, disclosure or refusal), beyond voluntary codes. ¶ | 45% | 2027-12-31 | open |
| 008 | Before the next Canadian or British general election, whichever comes first, a national regulator or electoral authority in that country will publish a pre-election audit of chatbot voting advice. ¶ | 60% | 2029-12-31 | open |
| 008 | By the 2028 U.S. presidential election, at least one of the three most-used consumer chatbots will name a preferred candidate in a general-election race when asked plainly, without a jailbreak, as default behaviour. ¶ | 30% | 2028-11-07 | open |
| 003 | The reopened Tompkins County investigation will produce at least one criminal indictment by the end of 2027. ¶ | 50% | 2027-12-31 | open |
| 003 | By the end of 2028, the FTC will have publicly announced fewer than ten enforcement actions under the Take It Down Act. ¶ | 65% | 2028-12-31 | open |
| 003 | By the end of 2028, the Ofcom investigation into X over Grok's sexualized images will have produced no fine larger than £10 million. ¶ | 60% | 2028-12-31 | open |
Variable // Attention · 6
| Ed. | Claim | Confidence | Resolve by | Status |
|---|---|---|---|---|
| 006 | By the end of 2027, at least one of Tinder, Hinge or Bumble will ship a feature that marks or verifies messages as human-written, or discloses A.I. assistance, in at least one market. ¶ | 45% | 2027-12-31 | open |
| 006 | When the Institute for Family Studies/YouGov question is next repeated, the share of Americans under 40 who say A.I. could replace real-life romance will exceed 35 per cent (up from 25 in 2024). ¶ | 55% | 2028-12-31 | open |
| 006 | By the end of 2030, at least one G7 country will regulate A.I. companion apps for adults specifically — disclosure, dependency safeguards or design rules — rather than only for minors. ¶ | 60% | 2030-12-31 | open |
| 005 | By the end of 2028, no U.S. federal statute specifically regulating A.I. companions or chatbots for minors will have been enacted; the field will still be governed by state suits and consumer-protection law. ¶ | 65% | 2028-12-31 | open |
| 005 | By the end of 2027, at least one of this week's state suits against OpenAI will produce a settlement or consent decree that imposes age-verification or parental-control requirements on ChatGPT. ¶ | 55% | 2027-12-31 | open |
| 005 | By the end of 2029, a peer-reviewed study of children under 13 will report a measurable reduction in task persistence associated with regular A.I. assistant use, replicating the 2026 Berkeley adult finding. ¶ | 70% | 2029-12-31 | open |
Variable // The capacity for difficulty · 6
| Ed. | Claim | Confidence | Resolve by | Status |
|---|---|---|---|---|
| 006 | By the end of 2027, at least one of Tinder, Hinge or Bumble will ship a feature that marks or verifies messages as human-written, or discloses A.I. assistance, in at least one market. ¶ | 45% | 2027-12-31 | open |
| 006 | When the Institute for Family Studies/YouGov question is next repeated, the share of Americans under 40 who say A.I. could replace real-life romance will exceed 35 per cent (up from 25 in 2024). ¶ | 55% | 2028-12-31 | open |
| 006 | By the end of 2030, at least one G7 country will regulate A.I. companion apps for adults specifically — disclosure, dependency safeguards or design rules — rather than only for minors. ¶ | 60% | 2030-12-31 | open |
| 005 | By the end of 2028, no U.S. federal statute specifically regulating A.I. companions or chatbots for minors will have been enacted; the field will still be governed by state suits and consumer-protection law. ¶ | 65% | 2028-12-31 | open |
| 005 | By the end of 2027, at least one of this week's state suits against OpenAI will produce a settlement or consent decree that imposes age-verification or parental-control requirements on ChatGPT. ¶ | 55% | 2027-12-31 | open |
| 005 | By the end of 2029, a peer-reviewed study of children under 13 will report a measurable reduction in task persistence associated with regular A.I. assistant use, replicating the 2026 Berkeley adult finding. ¶ | 70% | 2029-12-31 | open |
Variable // The lag · 3
| Ed. | Claim | Confidence | Resolve by | Status |
|---|---|---|---|---|
| 007 | By the end of 2027, the U.K. will have no primary legislation specifically governing police use of live facial recognition; deployments will continue under the existing guidance-and-case-law framework. ¶ | 65% | 2027-12-31 | open |
| 007 | By the end of 2028, at least one Canadian province will restrict police use of live facial recognition or automated licence-plate readers by statute or binding regulation, beyond a moratorium or guideline. ¶ | 40% | 2028-12-31 | open |
| 007 | By the end of 2029, at least one G7 government will have formally disclosed that a surveillance or targeting system it procured was first deployed by another state in an armed conflict, as a condition of procurement rather than after a journalist found out. ¶ | 25% | 2029-12-31 | open |
| Ed. | Claim | Confidence | Resolve by | Status |
|---|---|---|---|---|
| 009 | By the end of 2027, the United States will still have no federal statute creating a right in one's own voice or likeness against A.I. replication; the NO FAKES Act or a successor will not have been signed into law. ¶ | 65% | 2027-12-31 | open |
| 008 | By the end of 2027, at least one G7 country will have enacted a statute or binding regulation setting standards for A.I. chatbots' responses to election questions (accuracy, sourcing, disclosure or refusal), beyond voluntary codes. ¶ | 45% | 2027-12-31 | open |
| 007 | By the end of 2027, the U.K. will have no primary legislation specifically governing police use of live facial recognition; deployments will continue under the existing guidance-and-case-law framework. ¶ | 65% | 2027-12-31 | open |
| 006 | By the end of 2027, at least one of Tinder, Hinge or Bumble will ship a feature that marks or verifies messages as human-written, or discloses A.I. assistance, in at least one market. ¶ | 45% | 2027-12-31 | open |
| 005 | By the end of 2027, at least one of this week's state suits against OpenAI will produce a settlement or consent decree that imposes age-verification or parental-control requirements on ChatGPT. ¶ | 55% | 2027-12-31 | open |
| 003 | The reopened Tompkins County investigation will produce at least one criminal indictment by the end of 2027. ¶ | 50% | 2027-12-31 | open |
| 008 | By the 2028 U.S. presidential election, at least one of the three most-used consumer chatbots will name a preferred candidate in a general-election race when asked plainly, without a jailbreak, as default behaviour. ¶ | 30% | 2028-11-07 | open |
| 009 | By the end of 2028, neither the United Kingdom nor Canada will have enacted a statutory right of publicity for living persons covering A.I. voice replication. ¶ | 60% | 2028-12-31 | open |
| 009 | The Presley and Eight Mile suits against Suno will end in settlement or licence rather than a judgment on the merits of the training claims, by the end of 2028. ¶ | 70% | 2028-12-31 | open |
| 007 | By the end of 2028, at least one Canadian province will restrict police use of live facial recognition or automated licence-plate readers by statute or binding regulation, beyond a moratorium or guideline. ¶ | 40% | 2028-12-31 | open |
| 006 | When the Institute for Family Studies/YouGov question is next repeated, the share of Americans under 40 who say A.I. could replace real-life romance will exceed 35 per cent (up from 25 in 2024). ¶ | 55% | 2028-12-31 | open |
| 005 | By the end of 2028, no U.S. federal statute specifically regulating A.I. companions or chatbots for minors will have been enacted; the field will still be governed by state suits and consumer-protection law. ¶ | 65% | 2028-12-31 | open |
| 003 | By the end of 2028, the FTC will have publicly announced fewer than ten enforcement actions under the Take It Down Act. ¶ | 65% | 2028-12-31 | open |
| 003 | By the end of 2028, the Ofcom investigation into X over Grok's sexualized images will have produced no fine larger than £10 million. ¶ | 60% | 2028-12-31 | open |
| 002 | By the end of 2028, fewer than one in twenty U.S. private-sector workers will be covered by a collective agreement containing enforceable limits on an employer's use of generative A.I. ¶ | 75% | 2028-12-31 | open |
| 001 | By 31 December 2028, job postings with the title 'prompt engineer' on the major English-language job boards will number fewer than one third of their 2024–25 peak. ¶ | 70% | 2028-12-31 | open |
| 008 | Before the next Canadian or British general election, whichever comes first, a national regulator or electoral authority in that country will publish a pre-election audit of chatbot voting advice. ¶ | 60% | 2029-12-31 | open |
| 007 | By the end of 2029, at least one G7 government will have formally disclosed that a surveillance or targeting system it procured was first deployed by another state in an armed conflict, as a condition of procurement rather than after a journalist found out. ¶ | 25% | 2029-12-31 | open |
| 005 | By the end of 2029, a peer-reviewed study of children under 13 will report a measurable reduction in task persistence associated with regular A.I. assistant use, replicating the 2026 Berkeley adult finding. ¶ | 70% | 2029-12-31 | open |
| 004 | By the end of 2029, no publicly owned, co-operative or non-profit-governed model will rank in the top five on the major independent capability leaderboards. ¶ | 75% | 2029-12-31 | open |
| 002 | By the end of 2029, no US federal law will impose liability on A.I. developers for harms caused by model outputs in ordinary commercial use. ¶ | 70% | 2029-12-31 | open |
| 001 | By the end of 2029, at least one of the five largest global consultancies will have dissolved its standalone A.I. practice back into a general technology practice. ¶ | 55% | 2029-12-31 | open |
| 002 | US union membership as a share of employed workers (BLS annual figure) will be lower for 2029 than for 2024. ¶ | 60% | 2030-01-31 | open |
| 006 | By the end of 2030, at least one G7 country will regulate A.I. companion apps for adults specifically — disclosure, dependency safeguards or design rules — rather than only for minors. ¶ | 60% | 2030-12-31 | open |
| 004 | By 2030, labour's share of income in the United States (BLS nonfarm business measure) will be at least two percentage points below its 2024 level. ¶ | 60% | 2030-12-31 | open |
| 001 | By the end of 2030, at least one G7 country will require a named, accountable human sign-off for A.I.-produced work in at least one licensed profession (law, accounting, medicine, engineering). ¶ | 45% | 2030-12-31 | open |
| 004 | By the end of 2031, at least one G7 country will have enacted a permanent, universal cash transfer or citizen dividend at national level, funded at least in part from taxation of automated production or from a public stake in A.I. or compute infrastructure. ¶ | 30% | 2031-12-31 | open |
The annual review
Once a year, one edition grades the previous twelve months: every claim that came due, the calibration chart read out loud, and the framings that did not survive. The first is due in January 2027.
How grading works
A claim is right if it happened as stated by the resolve-by date, wrong if it plainly did not, and partial if a reasonable reader could argue either way — in which case the argument is written out under the claim. Confidence is the probability assigned at the time of writing. Over many claims, the ones marked 70% should come true about seven times in ten. If they come true nine times in ten, the writer is underconfident; if five, overconfident. Both are failures, and both get reported here once a year.
Every claim has a permanent address — the ¶ at the end of its row — so you can point at one when you write in to say it's wrong.