ai-law-tracker 0.1.0

Official Python SDK for the AI Law Tracker API — audited AI-regulation data (US state + federal, EU, global) in three lines.

AI Law Tracker — Python SDK

Official Python client for the AI Law Tracker API — the audited dataset of AI regulation across US state + federal, the EU, and global jurisdictions. Laws, news, changelogs, obligations, penalties, compliance assessment, deadlines, and webhooks.

pip install ai-law-tracker

Quick start (3 lines)

from ai_law_tracker import Client

alt = Client()                          # anonymous tier; no key required
laws = alt.laws.list(scope="federal")   # -> a Page of records

for law in laws:
    print(law["identifier"], "—", law["title"])

Authentication

The API has a free anonymous tier (rate-limited by IP). For higher limits, richer fields, and the changelog/webhook endpoints, use a free API key (no card required):

from ai_law_tracker import Client

# 1) Issue a free key by email (anonymous call is fine)
Client().account.create_key("[e-mail]")   # emails you a key

# 2) Use it — explicitly, or via the ALT_API_KEY env var
alt = Client(api_key="alt_live_...")
print(alt.account.get())   # your tier, limits, and live usage

Set ALT_API_KEY in your environment and Client() picks it up automatically. Paid tiers ($29 / $99 / $299 a month) raise quotas and unlock Pro-only endpoints — see https://ai-law-tracker.com/pricing.

Pagination

List endpoints return a Page you can iterate, index, and len(), plus .total, .limit, .offset, and .has_more:

page = alt.laws.list(scope="state", jurisdiction="colorado", limit=50)
print(page.total, "records match;", len(page), "on this page")

To sweep every matching record without managing offsets yourself, use .iterate(...) — it pages transparently and yields one record at a time:

for law in alt.laws.iterate(scope="eu", in_force=True):
    ...

for hit in alt.search.iterate("facial recognition", scope="state"):
    ...

Endpoints

Grouped as resources on the client:

Resource	Methods
alt.laws	list, iterate, get(id), history(id), sources(id), citations(id)
alt.search	query(q, ...), iterate(q, ...)
alt.news	list, iterate
alt.changes	list, iterate (poll since= for what changed)
alt.feed	list (lean recent-changes stream)
alt.jurisdictions	list, countries, states
alt.sectors	list, get(sector)
alt.compliance	obligations, penalties, assess, report
alt.deadlines	list (JSON or iCal via format="ical")
alt.bills	list, get(slug)
alt.webhooks	list, get(id), create, delete (Pro+)
alt.account	get, revoke, create_key(email), create_token(email)
top level	alt.health(), alt.accuracy(), alt.categories(), alt.openapi()

Examples:

# Poll the changelog for everything that moved since a timestamp
changes = alt.changes.list(since="2026-07-01T00:00:00Z", scope="federal")

# Compliance obligations for a sector + jurisdiction
obs = alt.compliance.obligations(jurisdiction="colorado", sector="hr")

# Risk assessment from a company profile
result = alt.compliance.assess(state="CA", sector="hr", aiUse="resume_screening")

# Upcoming compliance deadlines as an iCal feed (text/calendar)
ics = alt.deadlines.list(scope="eu", format="ical")

Error handling

Every non-2xx response raises a typed exception carrying .status, .code, .message, and .docs:

from ai_law_tracker import NotFoundError, RateLimitError, AuthenticationError

try:
    alt.laws.get("does-not-exist")
except NotFoundError as e:
    print(e.code, e.message)
except RateLimitError as e:
    print("retry after", e.retry_after, "seconds")

All inherit from ALTError. Transient failures (timeouts, 5xx, 429) on GET/DELETE are retried automatically with exponential backoff (configurable via Client(max_retries=...)).

Configuration

Client(
    api_key=None,        # or set ALT_API_KEY
    base_url="https://ai-law-tracker.com/api/v1",
    timeout=30.0,        # seconds
    max_retries=2,       # transient GET/DELETE retries
)

Client is a context manager (with Client() as alt: ...) and has .close() to release the connection pool.

Responsible use

This SDK is designed to be a well-behaved API citizen and does not provide any way to bypass the server's tier gating or rate limits:

• Requests are sequential with a single connection — no parallel fan-out or concurrent request flooding.
• Retries are conservative: a small default (max_retries=2), exponential backoff with jitter, and only on idempotent GET/DELETE. Writes/POSTs are never retried automatically.
• 429 responses and their Retry-After header are respected; a persisted rate limit surfaces as a RateLimitError (with .retry_after) rather than being retried away.
• iterate(...) pages one page at a time, in order, through the same rate-limited transport. It is a convenience for legitimate paging, not a bulk scraper — please honour your tier's quotas.

If you need high-volume or bulk access, use an appropriate paid tier or contact [e-mail] rather than working around the limits.

Links

• API docs & playground: https://ai-law-tracker.com/developers
• OpenAPI 3.1 spec: https://ai-law-tracker.com/api/v1/openapi.json
• Get a free key: https://ai-law-tracker.com
• Node/TypeScript SDK: ai-law-tracker on npm

License

MIT (SDK code). API data is CC BY 4.0 — attribution required, informational only, not legal advice.