The two largest patterns on the sky agree on a direction, and randomness says they should not. Decompose the CMB into multipoles and the quadrupole and octupole, the largest-scale temperature patterns the universe offers, turn out to be aligned with each other: their preferred axes point the same way to within about 10 degrees, a configuration dubbed the Axis of Evil (Land and Magueijo 2005) with chance probability well below a percent, sharpening to 2.8 sigma in combined analyses. The alignment survived the instrument change from WMAP to Planck untouched, ruling out the satellite-specific systematics that were the first suspicion, and it keeps company: the same general direction recurs in the hemispherical power asymmetry, the parity asymmetry, and the multipole-vector statistics that link the large-angle sky to the dipole and ecliptic.
ΛCDM's initial conditions are statistically isotropic by construction: every multipole's orientation is an independent random draw, so alignments carry no meaning and can only be accidents. The model's defenses are accordingly statistical rather than physical: a posteriori selection (the alignment was noticed, not predicted), look-elsewhere accounting, and the unanswerable observation that some configuration had to occur. These are legitimate cautions and unfalsifiable comforts; they explain nothing, predict nothing, and must be re-deployed for each member of the anomaly family separately, multiplying accidents the model cannot connect.
The standing is two decades of stubborn persistence: the alignment is cosmic-variance-limited in temperature and will never improve there, polarization offers the nearly independent retest (LiteBIRD, CMB-S4), and the question remains whether the largest scales of the universe are accidentally organized or organized by something the standard model lacks.